03
COMMISSION OF THE EUROPEAN COMMUNITIES
STUDY OF THE COMMUNÏTY BENEFIT OF A FIXED CHANNEL CROSSING
REV1SED MAIN REPORT
JANUARY 1980
COOPERS & LYBRAND ASSOCIATES LIMITED MANAGEMENT AND ECONOMIC CONSULTANTS 1/ i
TABLE OF CONTENTS Page 0 INTRODUCTION / 0.1 The Previous Channel Studies 0.1 0.2 Objectives of This Study 0.13
1 MODELLING TRAVEL BEHAVIOUR ACROSS THE CHANNEL j 1.1 Préface 1.1 1.1.1 Objectives 1.1 1.1.2 Division of Modelling Responsibility 1.2
1.2 The Models: UK Leisure Travel to Europe 1.3 1.2.1 Modelling UK Leisure Travel to Continental Europe 1.3 1.2.2 The Model of Demand for UK Leisure Trips Abroad 1.3 1.2.3 Geographical Distribution of UK Leisure Trip Demand 1.7 1.2.4 The Model of Trip Type 1.8 1.2.5 The Model of Destination Choice 1.9 1.2.6 The Model of Route Choice 1.10 1.2.7 The Route Choice Model for Car Travellers 1.11 1.2.8 The Choice between Ship and Hovercraft 1.14 1.2.9 The Route Choice Model for Independent Non-Car Travellers 1.15 1.2.10 The Route Choice Model for Package Travellers 1.16
1.3 The Models: the SETEC Model of Continental Leisure Travel to the UK 1.18
1.4 The Models: Business Travel Across the Channel 1.19 1.4.1 The Model of UK Business Travel Demand 1.19
1.5 The Models: Non-European Travel Across the Channel 1.23 1.5.1 Non-European Travel Demand 1.23 003
Page
1.6 The Models: Freight 1.27 1.6.1 The SETEC Model of Demand for Freight Travel 1.27 1.6.2 Route Choice for Freight Traffic 1.27 1.6.3 Patterns of Freight Movement in 1977 1.28 1.6.4 The Costs of Various Modes of Freight Transport 1.29 1.6.5 Ro-Ro Haulage Charges 1.30 1.6.6 Lo-Lo Container Haulage Charges 1.30 1.6.7 Road Vehicle Ro-Ro and Lo-Lo Container Sea Crossing Tariffs 1.31 1.6.8 Rail Haulage and Sea-Crossing Tariffs 1.31 1.6.9 Summary of Haulage Costs 1.33 1.6.10 Transit Times for Various Modes 1.33 1.6.11 Port and Sea-Crossing Times 1.35 1.6.12 The Model of Route Choice for Freight 1.36 1.6.13 The Application of the Freight Route Choice Models 1.37
2 THE BASE FORECASTS 2.1 Préface 2.1 2.1.1 Division of Responsibility 2.1 2.1.2 The Use of the Models in Forecasting 2.1 2.1.3 Developments in Passenger Traffic 1972-77 2.2
2.2 The Economie and Démographie Assumptions Made 2.5 2.2.1 Population Growth 2.5 2.2.2 Gross Domestic Product and Consumers1 Expenditure 2.5 2.2.3 Changes in Relative Priées 2.7 2.2.4 Other Exogenous Variables 2.9 2.2.5 Household Size 2.9 2.2.6 Car Ovmership 2.11 2.2.7 Travel Impédance in the UK 2.12 2.2.8 Friends and Relatives Abroad 2.12 2.2.9 Miscellaneous. 2.12 004
Page
2.3 The Base Forecast of UK Leisure Trips Abroad 2.13 2.3.1 The Base Forecast of Total Demand 2.13 2.3.2 Geographical Distribution cf UK Demand 2.14 2.3.3 Forecasts of UK Leisure Trip Demand by Type 2.16 2.3.4 Forecasts of UK Leisure Trip Demand by Destination 2.17
2.4 The Base Forecast of Continental Leisure Trips to the UK 2.18
2.5 Summary of Base Forecasts of Leisure Trips 2.19
2.6 The Base Forecast of the Demand for Business Trips 2.21 2.6.1 Business Trips by UK Résidents 2.21 2.6.2 Business Trips by Continental Résidents 2.24 2.6.3 Summary of the Forecasts of Business Trips 2.24
2.7 The Base Forecast of the Demand for Travel by Non-Europeans 2.26
2.8 The Base Forecast for Freight Traffic 2.27 2.8.1 Total Trade Flows 2.27 2.8.2 Unitised Trade Flows 2.27 2.8.3 The Forecast of Unitisation 2.28
3 ROUTE CHOICE IN THE ABSENCE OF A LINK 3.1 Cross-Channel Services in the Absence of a Link 3.1
3.2 UK Leisure ïraffic 3.4 3.2.1 Car Travellers 3.4 3.2.2 Independent Non-Car Travellers 3.4 3.2.3 Package Travellers 3.6
3.3 Continental Leisure Traffic SETEC
3.4 Business Traffic SETEC
3.5 Non-European Traffic SETEC
3.6 Freight 3.7 005
Page
4 THE ALTERNATIVES 4.1 Préface 4.1 4.1.1 The Major Alternatives 4.1 4.1.2 Structure of the Alternatives 4.1 4.1.3 The Alternatives Analysed 4.2 4.1.4 The Relevance of the Analysis to Other Alternatives 4.2
4.2 The Single-Track Rail Tunnel 4.3 4.2.1 Description: Capacity 4.3 4.2.2 Opération 4.3 4.2.3 Associated Capital Investment 4.4 4.2.4 Capital Costs 4.4 4.2.5 Services Offered 4.5
4.3 The Double-Track Rail Scheme 4.6 4.3.1 Description: Capacity 4.6 4.3.2 Opération 4.7 4.3.3 Associated Capital Investment 4.7 4.3.4 Capital Costs 4.8 4.3.5 Services Offered 4.10
4.4 The 'LINKINTOEUROPE' Road Bridge Scheme 4.11 4.4.1 Description: Capacity 4.11 4.4.2 Associated Capital Costs 4.11 4.4.3 Associated Capital Improvements 4.11 4.4.4 Services Offered 4.12
4.5 A Road Bridge and Single-Track Rail Combination 4.13
4.6 Other Fixed Link Schemes 4.14 4.6.1 Other Variants of the 'LINKINTOEUROPE' Bridge Scheme 4.14 4.6.2 The Euro-Bridge Scheme and Its Variants 4.14 4.6.3 The Channel Tunnel Island Project 4.15 4.6.4 Wimpey and Others: A Submerged-Tube Double-Track Tunnel 4.16 4.6.5 The Channel Bridge Group: Wide Span Bridge Project 4.16 006
Page
4.7 Other Technological Developments 4.17 4.7.1 Jetfoil 4.17 4.7.2 Airships 4.17 4.7.3 A Pipeline 4.18
4.8 Physical and Service Characteristics of a New Link 4.19
5 TARIFFS 5.1 Pricing Policy 5.1
5.2 Passenger Tariffs on the Link and on Compétitive Crossings 5.2
5.3 Freight Tariffs 5.3 5.3.1 Scénarios 5.3 5.3.2 Reactions Assurned 5.4
5.4 Taxation 5.8
6 ROUTE CHOICE IN THE PRESENCE OF A LINK 6.1 UK Leisure Trips to Europe 6.1 6.1.1 Car Trave11ers 6.1 6.1.2 Non-Car Travellers 6.2 6.1.3 Summary of UK Leisure Traffic on Links 6.3
6.2 Freight 6.5
6.3 Summary of Traffic on the Link 6.7
7 THE EFFECT OF A FIXED LINK ON EXISTING MODES AND OPERATORS 7.1 Ferry Operators 7.1 7.1.1 Approach 7.1 7.1.2 Calculation of Capital Cost of New Vessels 7.2 7.1.3 Operating Costs 7.6 7.1.4 Revenues 7.8 7.1.5 Summary 7.9
7.2 The Ports 7.12 7.2.1 Capital Investinent 7.12 7.2.2 Existing and Proposed Port Capacity 7.14 7.2.3 Forecast of Demand for Port Facilities 7.14 7.2.4 Loss of Port Revenue • 7.16 7.2.5 Réduction in Port Operating Costs 7.17 007
Page
7.3 Aviation 7.19 7.3.1 Airports 7.19 v7.3.2 Airlines 7.19
7.4 Railway Operators 7.21 7.4.1 Capital Costs 7.21 7.4.2 Passenger Revenues 7.21 7.4.3 Freight Revenues 7.22 7.4.4 Operating Costs 7.22 • 7.4.5 Continental Railways 7.23
7.5 Road Authorities 7.24 7.5.1 UK 7.24 7.5.2 Continental Europe 7.24
7.6 Road Freight Haulage Costs and Revenues 7.25
8 METHODS OF EVALUATION 8.1 The Approach 8.1 8.1.1 Indicated by the Terms of Référence 8.1 8.1.2 Contrast with a UK Transport Cost-Benefit Approach 8.1
8.2 Community Interest 8.2 8.2.1 Information for an EEC Décision 8.2 8.2.2 Community Policies 8.4 8.2.3 The Relevance of Cost-Benefit Analysis 8.4 8.2.4 The Différent Approaches to be Adopted 8.4
8.3 Summary 8.5
9 PROFITABILITY OF A FIXED LINK 9.1 Assumptions 9.1 9.1.1 Revenues and Costs Covered 9.1 9.1.2 Uncertainty 9.1 9.1.3 The Choice of Discount Rates 9.2 9.1.4 Tariffs 9.3 9.1.5 Inflation 9.3 008
Page
9.2 Capital Costs 9.3
9.3 Revenues 9.4
9.4 Operating Costs Formulae 9.5 9.4.1 Single-Track Tunnel 9.5 9.4.2 Double-Track Tunnel 9.7 9.4.3 Road Bridge 9.7 9.4.4 Road Bridge and Single Rail Combination 9.8
9.5 Operating Costs 9.8
9.6 Operating Surpluses 9.9
9.7 Profitability 1.10
9.8 Interprétation 9.11 9.8.1 The Tunnels 9.11 9.8.2 The Tunnels or the Bridge 9.12 9.8.3 The Single-Track Rail Facility and the Bridge 9.13
9.9 Tariff Sensitivity Tests 9.14 9.9.1 A Compétitive Response from the Ferries for Non-Car Passengers 9.14 .9.9.2 A Compétitive Response from the Ro-Ro Ferries 9.16
9.10 Conclusion 9.17
10 THE TRANSPORT COST-BENEFIT ANALYSIS 10.1 The Interests to be Considered 10.1
10.2 Choice of Discount Rates 10.2
10.3 User Benefits 10.3 10.3.1 Division of Responsibility 10.3 10.3.2 The Approach 10.3
10.4 Calculation of User Benefits 10.5 10.4.1 UK Leisure Travel to Europe 10.5 10.4.2 Continental Leisure Travel to the UK 10.8 10.4.3 UK and Continental Business Travel 10.9 10.4.4 Non-European Travel 10.9 10.4.5 Freight 10.11 009
Page
10.5 Summary of User Benefits 10.12
10.6 Impact on Ferry Operators 10.12 10.6.1 Capital Costs 10.13 10.6.2 Revenues 10.20 10.6.3 Operating Costs 10.20
10.7 Summary of Impact on Ferry Operators 10.21
10.8 Impact on UK ports 10.23 10.8.1 Capital Costs 10.23 10.8.2 Revenues 10.32 10.8.3 Operating Costs 10.35
10.9 Summary of Impact on UK Ports 10.36
10.10 Impact on Continental Ports 10.37
10.11 Impact on Airlines and Airports 10.37
10.12 Impact on UK Railway Operators 10.38 10.12.1 Capital Costs 10.38 10.12.2 Revenues 10.40 10.12.3 Operating Costs 10.41
10.13 Summary of Impact Impact on Railways 10.43
10.14 Impact on Continental Railways 10.44 10.14.1 Capital Costs 10.44 10.14.2 Operating Surpluses 10.45
. 10.15 Impact on UK Road Authorities 10.46 10.15.1 Kent 10.46 10.15.2 Greater London 10.48 10.15.3 Summary 10.49
10.16 Impact on UK Road Freight Operators 10.50
10.17 Impact on Continental Road Authorities and Road Haulage Operators 10.51
10.18 The Economie Returns 10.52
10.19 Interprétation 10.53 10.19.1 The Tunnels 10.53 10.19.2 The Tunnel and the Bridge 10.53 010
Page
11 EXTERNAL EFFECTS OF A FIXED LINK 11.1 Approach 11.1 11.1.1 The External Effects Considered 11.1 11.1.2 Irrelevance of Traffic Growth 11.2
11.2 The Measurement and Evaluation of Enviromental Impacts 11.5 11.2.1 Built Environment 11.5 11.2.2 Transport Noise 11.5 11.2.3 Road Noise 11.6 • 11.2.4 Aircraft Noise 11.8 11.2.5 Air Pollutants 11.9 11.2.6 Road Accidents 11.13
11.3 The Single-Track Tunnel 11.15 11.3.1 The Continental Side of the Channel 11.15 11.3.2 Kent: The View of the Council 11.15 11.3.3 London 11.17
11.4 The Double-Track Tunnel 11.19 11.4.1 Effect on the Continent 11.19 11.4.2 Kent: View of the County Council's Officers 11.19 11.4.3 London: the GLC Officers" Views 11.21
11.5 The Bridge 11.22 11.5.1 The Continent 11.22 11.5.2 Kent 11.23 11.5.3 London 11.23 11.5.4 The Bridge Itself 11.23
11.6 A Submerged Tube Project 11.25 11.6.1 Kent: the County Council Officers' Views 11.25 11.6.2 The British Consultants' Views 11.26
11.7 The British Consultants1 View 11.27 11.7.1 Road Noise 11.27 11.7.2 Aircraft Noise 11.30 11.7.3 Air Pollution 11.31 11.7.4 Road Accidents 11.32 11.7.5 Summary. 11.33 011
Page
12 DISTRIBUTIONAL IMPACTS 12.1 Approach 12.1
12.2 The Distribution of Costs and Benefits Among Transport Users 12.2 12.2.1 The Approach 12.2 12.2.2 The Distributional Impact of the Transport Cost-Benefit Study 12.2
12.3 The Distribution of Benefit between Régions 12.4 12.3.1 Approach 12.4 12.3.2 Potential Impacts 12.4 12.3.4 Diversion of Funds 12.7 12.3.5 Construction Phase Effects 12.8 12.3.6 Operating Phase Effects 12.9 12.3.7 Impact on the Pattern of Trade and Passenger Journeys 12.10 12.3.8 Investment Décisions 12.10 12.3.9 Evaluation of Régional Impact 12.11
12.4 The Distribution of Benefit between Sectors 12.20 12.4.1 Approach 12.20 12.4.2 The Multiplier Analysis 12.21 12.4.3 The Concept of Multipliers 12.21 12.4.4 General Description of the Multiregional Multiplier Model 12.23 12.4.5 Demand for and Supply of Commodities 12.23 12.4.6 Household Consumption and Income Génération 12.25 12.4.7 Modelling intra-EEC Trade Flows 12.26 12.4.8 A Summary of the Multiplier Model 12.27 12.4.9 Data Requirements and Detailed Model Spécification 12.29 12.4.10 Strategy for Model Construction 12.32 12.4.11 Calculation of Model Parameters 12.33 12.4.12 Final Demand Indices 12.34 12.4.13 Gross Output Indices 12.35 12.4.14 Change in the Exchange Rate and the EUA 12.36 012
Page 12.5 Evaluation of the Sectoral Impact of the Link 12.38 12.5.1 The Multiplier Impact of Link Construction 12.38 12.5.2 The Total Multiplier Impact 12.40 12.5.3 The Impact of Link Construction on Energy Use 12.40 12.5.4 The Total Impact on Energy Use 12.40
13 THE BENEFIT TO THE COMMUNITY 13.1 Approach 13.1
13.2 Relevance of the Community1s Policies 13.1 13.2.1 Economie Policies 13.1 13.2.2 The Community1s Steel Policy 13.2 13.2.3 Compétition Policy 13.3 13.2.4 Régional Policy 13.4 13.2.5 The Community1s Environmental Policy 13.4 13.2.6 Transport Policy 13.7 13.2.7 The Community's Energy Policy 13.10
13.3 European Cost-Benefit Study 13.11
14 TESTING THE ROBUSTNESS OF THE RESULTS 14.1 Recommendations on Further Work 14.1 14.1.1 Sensitivity Tests 14.1 14.1.2 The Schemes 14.2 14.1.3 Scheme Variants 14.3
14.2 Conclusions 14.5
\ 013
n U u 014
H 0.1 THE PREVIOUS CHANNEL STUDIES
There has, over the years, been a large number of studies of ] differing depth and complexity concerned with a fixed Channel crossing. The most important récent published work prior to those with which Coopers & \ I Lybrand were concerned was the "Proposais for a Fixed Channel Link" published by HMSO (Cmnd. 2137) in September 1963. This was a report of a Working Group ~v of British and French officiais which examined thé advantages of a Channel u Tunnel compared with a Channel Bridge. The study found that a rail tunnel -^ would provide a better économie return than a bridge. The expected économie j rate of return varied between 7.4% and 14.4% depending upon the assumptions made as to the level of traffic using the Tunnel. il (J In February 1964, the Minister of Transport announced that as a resuit of studies jointly undertaken, the British and French Governments had / ' agreed that the construction of a rail Channel Tunnel was technically possible, and that in économie terms it would represent a sound investment of the two j I countries1 resources. Subject to further discussion of the légal and financial problems involved, the two Governments had therefore decided to proceed with \ j the project. In July 1966 the British and French Prime Ministers announced î. ' that, "subject to finding a solution for the construction work on mutually ['î acceptable terms, the two Governments hâve now taken a décision that the Tunnel t ; should be built". In October 1966 it was announced that the Tunnel would be financed by a group of private financial interests which would also manage Its construction. The completed Tunnel would then be handed over to an Anglo-French operating body who would be responsible for payments of capital and interest to the private group from the Tunnel revenues- After prolonged negotiations between various financial groups, the British Channel Tunnel Company (BCTC) and the Société Française du Tunnel sous la Manche (SFTM) were formed. Their proposais to finance the programme of , studies and eventual construction of the Tunnel were accepted by the British and French Governments in March 1971.
: Rio Tinto Zinc Development Enterprises (RTZDE) were appointed Project Managers to the BCTC and Situmer were Project Managers to SFTM.
' The Preliminary Phase Studies
The Companies agreed with the Government to carry out between May 1971 and May 1972 Preliminary Studies, at the end of which the main features of the Project should be definitely settled and the décision made to undertake 015 o.:
the Final Studies and the Initial Phase of the works with a view to starting the final works in as short a time as possible. The Preliminary Studies covered both Technical and Economie Studies. The brief for the Economie Studies was:-
(i) to perfect the methodology to be used both for surveys and for the drawing-up of the analyses; (ii) to carry out a pilot survey at the frontier of passengers travelling by air and by sea; (iii) to carry out a pilot home interview survey of individuals taking holidays abroad and on reasons for their journeys; (iv) to analyse available documents having useful information about goods likely to be transported via the tunnel; (v) to carry out a pilot survey of freight traffic; (vi) to participate in the pursuit and évaluation of ancillary revenue from which the Tunnel Concessionary Company would profit; (vii) according to the results of ail the technical and économie studies outlined above, to update the cost-benefit analysis and the profit and risk analysis for the project; (viii) finally, to propose the detailed contents and the programme for the Economie Studies to be completed during the next period, that of the Final Studies.
The respective Project Managements commissioned Coopers & Lybrand Associates Limited and SETEC-Economie to advise them on the économie and financial aspects of the work. The Consultants were set two main objectives in the Preliminary Phase. The principal objective was "te prépare with respect to both the data collection and at the methodological level, for the économie studies to be undertaken in the following phase", referred to as Phase I, but at the same tirae the opportunity was taken in the Preliminary Phase to update the profit and risk analysis and the cost-benefit analysis with respect to the project.
The work carried out in the Preliminary Phase is recorded in the document produced by Situmer and RTZDE entitled "Channel Tunnel Main Report, April 1972". 016 0.3
Phase I Studies
The Phase I Studies were carried out between September 1972 and May 1973» The Companies reappointed Coopers & Lybrand (C & L) and SETEC-Economie and agreed a programme of work which allowed each consultant to take maximum advantage of its work in the Preliminary Phase. It was agreed that each consultant would hâve spécifie responsibilities for forecasting particular éléments of the total demand and diversion to the Tunnel. The division of responsibilities was as follows:-
Traffic Category Total Demand Diversion to Tunnel
Briti'sh Holidaymakers C & L C & L Continental Holidaymakers SETEC SETEC UK Business Travellers C & L SETEC Continental Business Travellers SETEC SETEC Non-European Travellers C & L SETEC Freight SETEC C & L
The work on ferry costs and charges and the financial analyais was carried out jointly by the two consultants. Each consultant produced separate fleet scheduling models.
The Phase I Studies are summarised in the "Channel Tunnel Summary Report, June 1973" produced by RTZDE and Situmer. The Economie Studies are described in more détail in the "Channel Tunnel Economie Report" produced at the same date. Both consultants* work is summarised in the Main Summary of the Economie Report. Separate reports were then prepared by each consultant describing in détail the methods used for those areas of the work for which they were responsible.
The portions of the work for which Coopers & Lybrand were responsible are summarised below. The "Passenger Studies, Main Report" gives a very full description of the raethod used to prépare thoae passenger forecasts for which CAL were re3ponsible. It reviewed developments in the British holidaymaker market over the period 1962-71, then described the surveys carried out for the studies. In addition to the Depth Interview Survey, Household Survey and IPS Trailer Survey which were commissioned in the Preliminary Phase the following additional surveys were carried out:- 01? 0.4
(a) The Non-Abroad Holidaymaker Survey, examining the reasons why sorae raembers of the adult population had never ^j taken holidays abroad. A séries of group discussions followed by structured interviews with a représentative 1 sample of 200 adults was carried out to probe the reasons why sorae people had never partaken of holidays | / abroad. (b) The Random Individual Survey. A total of 4,546 interviews ; • were carried out with households. The main information ! obtained was detailed of ail holidays taken in the p'revious twelve months which could be cross-classified ' •> to the socio-economic characteristics of the households concerned. • (c) The Short Holidaymaker Survey. A total of 324 adults who had taken a short holiday abroad in the period 1970 i j to 1972 were interviewed. (d) Car Holidaymaker Survey. After the completion of the |~" Preliminary Phase when the very considérable importance t~-' of accompanied car holidaymakers to total Tunnel revenue -••) prospects was apparent, it was decided to undertake a l : spécial survey to examine in more détail the characteristics and behaviour of this group of travellers. A total of 3,250 postal questionnaires were returned giving détails of:- •• (i) the reasons for choice of cross-Channel routes; (ii) brief détails of holidays abroad over the period 1963-1972; • . (iii) extensive factual détails on the dimensions of the vehicles used for holidays abroad; (iv) attitudes to short holidays abroad; (v) socio-economic characteristics of the household. (e) Tunnel Reactions Survey. A total of 16 group discussions were held with 110 respondents to help assess attitudes to travel through the Tunnel. OIS
The Report summarises the results of thèse surveys. It then describes in détail the models and the exogenous assumptions used to produce forecasts for British leisure travel abroad, for British business travel abroad, and for non-European cross-Channel tr'avellers, and the resulting forecsts. For British leisure travellers, models are also described, and forecasts produced, for the choice of Channel crossing in the présence of a Tunnel.
The work carried out to estimate the division of freight traffic to the tunnel is described in the "Freight Studies, Main Report". For this work a number of surveys were undertaken. The Spécial Commodities Survey was concerned essentially with those conunodities which were not considered unitisable but which raight use the Tunnel in full train loads. The Export Unitisable Survey suppleraented and expanded upon the previous work on the pattern of UK exports carried out in the Preliminary Phase. The Import Unitisable survey replaced the pilot survey of British imports carried out at the time of the preliminary studies. A total of 412 interviews were carried out with importers. In addition, an extensive survey of costs and charges was carried out, the information from which was also used in the Fleet Scheduling Model (see below). From thèse data, models were developed to predict the choice of mode (roll-on/roll-off or container) and route for unitisable freight, and applied, using assumptions about future services.
C & L and SETEC prepared différent Fleet Scheduling Models which were used to estimate the effect on the operators of cross-Channel ferries of the forecast changes occurring as a conséquence of the Tunnel.
Given the forecasts produced, a financial analysis was undertaken. It was shown that on the basis of the low estimâtes of the growth in GDP then the financial internai rate of return on the Tunnel project was expected to be 14 per cent. Various sensitivity tests were also carried out and it was shown:-
(a) that the profitability of the Tunnel is relatively robust to major increases in construction costs or decreases in revenues: in the most adverse case considered, the return exceeded 10 per cent; (b) operating profits were likely to be sufficient to cover loan interest payments except only in extrême and (in the Consultants' view) unlikely circumstances. 019
The United Kingdom Co3t-Benefit Study The Phase I studies, unlike the Preliminary Phase, did not include a cost-benefit analysis of the Tunnel project. During the course of Phase I the UK Department of the Environment decided that it was necessary to produce a transport cost-benefit study. Coopers & Lybrand were cornraissioned to carry out the study which considered the impact of the Channel Tunnel on transport costs and benefits to United Kingdom interests as compared with the continued development of the existing air and shipping services. In commissioning the study, the Department of the Environment directed that the consultants were to confine their attention to transport costs and benefits to the United Kingdom; other potential costs and benefits ari3ing from the environmental impact of the Tunnel and its effects on régional employment and relative incomes or on the UK balance of payments were specifically excluded from the study.
The work carried out for this study is described in the report "The Channel Tunnel: A United K:".ngdom Transport Cost-Benefit Study", published by HMSO in May 1973. It was shown that on the basis of the assumptions adopted "the transport costs and benefits of thé Tunnel discounted at 10 per cent per annum to 1973 gave a net présent value to the Tunnel of £292 million which was équivalent to an internai rate of return of 17.6 per cent". The corresponding figures based on the low forecasts were £148 million and 14.6 per cent. A range of sensitivity tests was also carried out in order to test the sensitivity of the Tunnel cost-benefit. It was found that the rate of return was relatively robust to even marked changes in the basic assumptions on which the forecasts had been based.
The cost-benefit report was not surprisingly the subject of some controversy. The main assumptions in the report which were questioned were:-
(a) the treatment of hovercraft; (b) the size of the shipping fleet required in the without Tunnel situation; (c) the assumptions on the relative tariffs that could be chargea by the shipping operators and the effect this would hâve on the percentage of the traffic using the Tunnel.
Thèse points were considered in more détail in the Phase II Cost-Benefit Studies. 020 o.7
The Channel Tunnel: Its Economie and Social Impact on Kent A study on the Economie and Social Impact of the Channel Tunnel in Kent was prepared by Economie Consultants Limited and published by HMSO in April 1973* The study concluded that, apart from an employment problem in Dover associated with a décline in the level of opérations of the existing sea ferries, the Tunnel would hâve little social and économie impact on Kent unless large scale industrial development was permitted by the planning authorities.
The Channel Tunnel: (Cmnd. 5430) This white paper was presented by the Secretary of State for the Environment- to Parliament in September 1973 in order to set out the UK Government's view of the project. It was concluded "that a Channel Tunnel would accord with our national interest. It would be the cheapest and most satisfactory way in the long run of providing for the dramatic and continuing increase in cross-Channel traffic and would reduce the barrier the Channel présents to our trade". It was also stated that "a high nuality rail link between the Tunnel portai and London with provision for through services to provincial' centres is essential to the success of the project". The rail link was expected to cost about £120 million in 1973 prices. It was therefore decided to proceed with Phase II of the project.
The Monopolies Commission Report: Cross-Channel Car Ferry Services
In April 1974 the Monopolies Commission presented their report on Cross-Channel Car Ferry Services. This report examined the opérations of the four British Cross-Channel operators who were:-
(a) the British Railways Board; (b) European Ferries Limited; (c) Southern Ferries Limited; (d) Hoverlloyd Limited. The report includes a detailed analysis of the profitability of thèse operators related to the pattern of service which was offered in 1973»
It was concluded that the monopoly positions of British Rail and European Ferries did not operate against the public interest. It was however stated that "the participation by cross-Channel ferry operators in collectively agreeing through the Harmonisation Conférence to charge fares at not less than coramon minimum rates and their participation in pooling agreements operate and may be expected to operate against the public interest". 021 0.8
The implications of the seasonal variability of deraand were examined 1 in the report. It was stated that "It is the opérators' policy to provide capacïty to meet virtually the highest level of demand likely to occur on any day, while charging the same fares on days of maximum demand as on any other days during the peak season". The Commission considered that in their view "this policy contributes more than any other factor to the high costs of opération, particularly on the Dover Straits routes". The policy also leads "to the provision of a larger number of ships than would otherwise be needed and this résulta in a low average utilisation of ships throughout the year". It was the Commission1s view that "the volume of resources involved in the provision of the services could be materially reduced by modification of the fare structure within the summer peak period".
Phase II Studies
The Phase II économie studies continued the séries of Channel Tunnel Studies commenced in 1971 with the Preliminary Phase, followed by the Phase I studies which were corapleted by June 1973. The Phase II studies were carried out over the period from January 1974 to the cancellation of the Channel Tunnel Project in January 1975.
It was considered unnecessary to repeat the surveys carried out in Phase I or develop new models so soon after the completion of Phase.I. The scope of the Phase II studies was therefore narrower than the previous phase, consisting primarily of:-
(a) sensitivity tests of the Phase I forecasts to changes in certain basic assumptions; (b) updating the statistical basis used in the previous phases to take account of the most récent information available to produce new traffic forecasts; (c) a more detailed study of the compétitive position of the Tunnel in relation to the cross-Channel ferries.
The form of the models developed in Phase I was retained, with the exception of the compétition study, for which new modela-were developed. The scope of the work to be undertaken was set out in détail in a contract between the British and French Channel Tunnel Companies and the Consultants, Coopers & Lybrand Associates Limited and SETEC-Economie, dated July 1974* The extent of the work carried out is 3ummarised below. - 022 0.9
Sensitivity Tests and Updating
In order to ensure that the Phase I models produced valid forecasts, various tests were undertaken in Phase II under alternative assumptions to indicate the sensitivity of the Phase I forecasts to the following changes:-
(a) the impact of the changes in transport costs arising from the increased price of energy; (b) the effect of an increase in air journey times of half-an-hour to allow for increased congestion and security checks at airports; (c) the effect of the development of a high-speed rail network in Western Europe; (d) the effect of lower rates of économie growth.
The sensitivity tests were carried out for the Phase I forecasts based on low GNP growth rates, for 1980 and 1990. The effect of the introduction of a motorail service through the Tunnel was also examined.
The updating of the Phase I forecasts occupied a large part of the Phase II studies. This was necessary, partly because of the passage of time (the Phase I forecasts were based on 1970 and 1971), but mainly because of the considérable change in expectations of growth in ail fields following the rise in energy priées.
The first requirement was therefore to reassess the économie growth assumptions used in the previous forecasts, in the light of the fuel price rises, and to propose, new central case assumptions for study purposes. The next step was to examine the most récent trends in traffic and revise the data base to take account of the latest available information. At the same time, the characteristics of the traffic were examined in as much depth as possible in order to détermine whether or not significant changes had occurred that might necessitate recalibration or reformulation of the models. However, it was not considered necessary or practicable to repeat the extensive surveys carried out in Phase I. The inputs to the demand modela were updated to 1973 or 1974 values, depending upon the latest information available.
Compétition Study
The development of a new model to study the compétitive position between the Channel ferries and the Tunnel was the main innovation in the Phase II studies. The purpose of the compétition study was to estimate more precisely the traffic that might be carried on the Tunnel, or ferries, at différent tariff levels. To do so it was necessary to devise a model which could simulate ferry opérations under alternative commercial policies. 023 0.10
In January 1975, the studies werë cancelled as a resuit of the Tunnel project being shelved by the Governments. Because of this, the major innovative features of the Phase II Studies were not brought to corapletion.
Provision of Ferry Trains
At the same time as the Phase II studies were proceeding, Coopers à Lybrand were requested by project management to undertake an examination of the number of ferry trains which would be needed to carry vehicles through the Tunnel. The work extended the studies of traffic peaking, prepared by Situmer in 1973 and 1974 in an extensive re-analysis of the Phase I survey data. Future trends in passenger and freight vehicle traffic peaking were predicted frora a considération of the factors affecting monthly, daily and hourly peaking characteristics. A simulation model was developed to détermine the optimum number of train sets required to meet specified service level criteria and a report on this work was presented to RTZDE in August 1974.
Phase II Cost-Benefit Study
Coopers & Lybrand were commissioned to repeat the Phase I cost-benefit studies. We were however, specifically requested to give attention to:-
(a) fleet size in the 'no Tunnel' situation, including the effect of peak pricing, alternative shipping technology and hovercraft; (b) port and harbour infrastructure and operating costs; (c) air, road and rail transport costs; (d) revision of the route/mode assignment network for passengers and freight in the 'no Tunnel1 situation; (e) a clear présentation of transport costs and benefits in resource ternis.
At the time of the cancellation of the Channel Tunnel Project we had just prepared a provisional estimate of the Tunnel costs and benefits. We were asked to continue with this work and to give advice on other topics of interest to the Channel Tunnel Adivsory Group which had been set up under the chairmanship of Sir Alec Cairncross (the Cairncross Committee). The work carried out was presented to the Department of the Environment as the "Channel Tunnel Provisional Estimâtes of Costs and Benefits", dated 18th April 1975. This gave, on the basis of the Preliminary.Phase II forecasts, our best estimate of the overall transport costs and benefits to UK interests from the Tunnel. It was estimated that if the preliminary traffic forecasts from 026
Phase II were accepted, then the Tunnel project would hâve a positive net présent value of £176 million at a 10 per cent discount rate. The NPV happened to be almost the same whether the Low Cost or the Intennediate Cost Strategy was adopted to provide rail connections from the Tunnel to London.
The Channel Tunnel and Alternative Cro3S-Channel Services (The Cairncross Report)"
The Channel Tunnel Advisory Group was set up in the suramer of 1974 by the Secretary of State for the Environment to détermine whether "the various économie and financial studies, underway and planned, provided an adéquate basis for deciding whether the Tunnel was in broad économie and commercial terms more advantageous to the UK than continued reliance on air and sea services and, if not, how thèse studies should be supplemented". This group, under the chairmanship of Sir Alec Cairncross, was asked to continue its work after the cancellation of the Channel Tunnel project and to report both on the original project and on the method of assessment to be ussd should such a venture be reviewed in the future. The report was published in July 1975 and comprised:-
(a) review of cross-Channel traffic and the alternative ways in which future growth could be handled; (b) discussion of key issues concerned with the size of the shipping fleet, the role.of hovercraft, and the UK rail link alternatives; (c) various présentations of the physical resource coats and cost-benefit implications of the project; (d) conclusions as to the studies undertaken, the project itself, and the co-ordination and planning issues raised by the project1s abandonnant.
Channel Tunnel Rail Link Intermediate Stops Study This study was carried out by Coopers & Lybrand Associates Limited for a working party of officers of the Greater London Council, Kent County Council, Surrey County Council and the London Borough of Croydon. The report was completed in March 1975. In the study the route allocation model of British holidaymaker cross-Channel traffic originally prepared for the Phase I Channel Tunnel Studies was modified so that it could be used to prépare forecasts of the potential passenger demand at alternative intermediate stops on the proposed high speed rail link between London and the Channel Tunnel. A separate model for the allocation of British business tripa was also developed. 025 0.12 n U It was shown that there was a considérable potential demand to use an out-of-town station at either East Croydon, on the M25 near Oxted, or at ,' j Tonbridge. r, Forecasts of Cro3s-Channel Road Traffic Through Kent Ports i i —~—~———^————————————————^—^———^————^——— 1 ' , This report concerned with the future level of road traffic using the > t cross-Channel freight and passenger services operating frora the Kent ports was [ j prepared for the General Planning Highways Division of the Department of the Environment between April and September 1975« The forecasts were presented ,' separately for 1981, 1986 and 1991 and 1996. A route allocation model based on the Phase I model for British holidaymakers, but with some minor amendments, ; was used to allocate the passenger traffic to alternative routes and the effect i of port capacity constraints on both freight and passenger traffic at each of the Kent ports was then considered. The peaking characteristics of passenger • and freight traffic were also examined.
t s • J 026 0.13
0.2 OBJECTIVES OF THIS STUDY
On 29th December 1978, the Commission of the European Communities commissioned Coopers & Lybrand Associates Limited to undertake its share of a new study of the feasibility of a Channel Link. It was to be done in conjunction with SETEC of France. The division of work between the parties is described subsequently. This report is the Final Report provided for in the Contract of that date. .
Annex I to the Contract specified two objectives:-
(a) to provide the commission with practical guidelines for the measurement of "Community interest" in major transport infrastructure projects. This should involve a considération both in theory and practice of the nature of such benefit, and possibly costs, and how they might be assessed and quantified; and (b) to provide a practical démonstration of the identification and measurement of Community interest by bringing up to date and re-assessing the case for a fixed link across the Channel between the UK and France. The case study is to be used as the example of the developraent of the concept of Community interest.
What distinguishes the study from the earlier ones discussed in 0.1 is:-
(a) the eraphasis given to the conceptualisation of Community benefit; (b) the need to revise ail forecasts given the passing of time. This has been made particularly important by the combination of économie recession and inflation; the very great increases in energy costs that hâve occurred and may be predicted to occur in future; and Britain's entry into the Common Market; (c) the need to revise costs; and (d) the much greater number of alternatives the Study is required to consider, treating each on its merits. It is also important to recognise what the study has been expected to achieve in another sensé. The alternatives vary greatly in the extent to which they hâve been designed, their technical feasibility has been tested, and they hâve been costed in détail. Neither indeed was it part of the Consultants1 remit to undertake an engineering or costing évaluation of the alternatives. 028
1
MODELLING TRAVEL BEHAVIOUR ACROSS THE CHANNEL 02S î.i
1.1 Préface
1.1.1 Objectives
To evaluate the impact of any major change in transport infrastructure, one must define the patterns of travel likely to develop first with and then without the change. To do thia, one needs to examine past and présent patterns of travel, and to dérive from them models of travel behaviour which explain adequately the factors affecting that behaviour and therefore why people behave as they do, both as passengers and as consignors of freight. Thèse will enable us to predict travel patterns in futures both with and without the'new transport development.
A. fixed link across the Channel alters the problem from that most frequently encountered in studies of this kind, in that the proposed development is not a replacement or an improvement of existing linka or a new link of an existing mode, but an alternative to existing msthods of transport (sea and air). One needs therefore to analyse the influence of thé characteristics of existing modes and routes on travel behaviour and compare thera with the corresponding characteristics of the new mode. This being so, the choice which it is most crucial to understand is the choice between routes. If we can adequately explain why a given traveller or consignor of goods chooses one route across the Channel rather than another, then we are likely to be able to predict their responses to a new link, provided we hâve described the routes in terms of carefully chosen relevant characteristics.
At présent, the traveller (or freight consignor) between Great Britain and Continental Europe has a wide choice of possible routes open to him; this range of choice has increased greatly in récent years, and is still increasing. There is no a priori reason why a new link could not provide an alternative to any of thèse routes, although, of course, a new link across the Straits of Dover will be most compétitive with the existing routes across the Straits.
Sea routes to France leave from many South coast ports, and to Belgium and the Netherlands from several East coast ports. In addition, there are longer ferry routes direct to Scandinavian, German, Spanish, and Portuguese ports. London is connected by air with a large number of Continental destinations, and, increasingly, so are Manchester and other airports outside London. The patterns of travel observed on thèse routes, and how they hâve developed, are described in Appendix A. 030
1.1.2 Division of Modelling Responsibility
The models of travel behaviour were separately derived for différent catégories of traveller, some by Coopéra & Lybrand and others by SETEC. Table 1.1.1 shows how their responsibilities were divided.
Table 1 .1 .1 ; Division of Modelling Responsibility
Total Demand Route Choice
UK leisure C.«l C & L Continental leisure SETEC SETEC UK business C & L SETEC Continental business SETEC SETEC Non-Europeans C & L SETEC Freight SETEC C & L
The models used by Coopers & Lybrand are described in subséquent sections. 0 031 i.3
L 1*2 The Models : UK Leisure Travel to Europe
'-j 1.2.1 Modelling UK Leisure Travel to Continental Europe
In previou3 Channel Tunnel studies, surveys of people's attitudes to j j holidays abroad suggested that it would be appropriate to model décisions -I about trips abroad in four stages. We hâve retained this structure, the four r-, stages being:-
L-i Stage 1 : Trip Demand --, The probability of taking a leisure trip abroad is assumed to be a [J function of certain socio-economic characteristics of an individual. Similarly the division of this demand between UK régions is assumed . J to dépend on average characteristics of the population of each région• Stage 2: Trip Type : The trips abroad arising in a given région are divided into three catégories: accompanied car trips, independent non-car trips, and "*";• package non-car trips. Their division between régions again dépends on characteristics of the population of each région. r- Stage 3: Destination Choice ;__ The proportion of trips to each country abroad in a given category is assuraed primarily to change in response to variation in the relative cost of living. Stage 4: Route Choice Trips from the UK to countries abroad were assigned to the various crossings using a perturbed least generalised cost method. The behaviour of travellers at each of thèse stages was analysed using three main sources of information:-
1. Surveys conducted for the previous Channel Tunnel Study; 2. Information published by the UK Department of Trade from the International Passenger Survey (IPS); 3. A detailed analysis of 1977 moveraents as revealed by the IPS. At every stage, many variables and, in some cases, a number of functional forms were examined. The final équations were chosen as statistically the most significant, while retaining behavioural plausibility.
1.2.2 The Model of Demand for UK Leisure Trips to Europe
The data for the estimation of this model consist principally of a file of surveyed individuals on which are stored, for each individual, the • number of leisure trips abroad taken in 1971, and a large number of socio- economic and geographical characteristics. Unfortunately, the information on ! the zone of résidence of thèse individuals wa3 no longer usable because of the 032 1.4
local government reorganisation of 1974, so the allocation of trip-makera to UK zones of résidence was modelled at a more aggregated level using the IPS tabulations of trips for 1 977*
As a first stage, a linear model was fitted, with the number of trips abroad as the dépendent variable, being explained as a linear function of the socio-économie characteristics. The form of the most suitable linear function is shown in Table 1.2.1. With seven independent explanatory variables, it explains 13.2$ of the variance. It should be noted that the dépendent variable hère is a discrète variable (the number of leisure trips abroad taken by a household can take the values 0,1,2,3,4, etc) and thus the proportion of \"5»2% of the variance explained with a continuous explanatory function is quite satisfactory.
Table 1.2.1: Linear Model for Total Holiday Demand
Contribution Variable Coefficient to r2
intercept -0.1721 household income (thousand £) 0.0614 0.0598 friends & relatives abroad(a' 0.1051 0.0346 number in household -0.0364 0.0144 âge of completion of éducation^' 0.0294 0.0086 UK travel impédance -0.0001609 0.0062 (équivalent minutes) social grade of respondent 0.0361 0.0048 number of cars in household^0' 0.0359 0.0040
0.1324
Notes: (a) This variable was defined to hâve the values: 1: No friends or relatives abroad 2: Friends or relating only outside Western Europe 3: Friends or relatives in Western Europe (b) This variable was defined to hâve the values: 1: 14 years 2: 15 years 3: 16 or 17 years 4: 18-20 years 5: 21 years or older (c) This variable was defined to hâve the values: 1 : . no cars 2: one car 3: two cars 4: three or more cars. 033 1.5
The relative importance of the variables can be seen from their marginal contribution to the proportion of the variance explained, given in the Table. The most important déterminant of pfopensity to take a trip abroad is household income. In addition, households with friends and relatives abroad are more likely to go abroad than those without, and larger households are less likely to go abroad than smaller. Households in parts of the country remote from the most frequently used ports and airports are less likely to go abroad than those close to thera, and car-ovming households are more likely to take a trip abroad than those without a car. The variables "âge of completion of éducation" and "social grade", while highly correlated with income, show an additional contribution to the observed cross-sectional variation in propensity. However, we feel that it would be unwise to use projected changes in thèse two variables in our forecasts, for reasons described below in Chapter 2. The variables in the régression were ail those recorded in the survey, with the exception of "UK travel impédance". This variable was inferred from the "mean egreas distance" variable, which was the weighted average distance of the respondent's zone from the ports and airports, and the impédances associated with time and distance by the route choice model, suitably deflated to 1971 values. The formula used was
J UK travel impédance = egress distance x ( 1 + cost per unit distance) speed value of time 1 The impédance is forraulated as a generalised cost, and expressed in équivalent minutes. This is necessary when using this variable to predict in ! the future. If impédance were expressed as a generalised cost in money units j then, as incomes rosé, the value of time, and hence the cost of the average } journey, would rise. This would lead, ail other things being equal, to the implausible conclusion that as incomes rise, average trip lengths would fall. i ' The generally accepted way round this difficulty is to express the impédance in i units of équivalent time. The first tenu represents the actual time taken for . the journey, and the second the cost of the journey, expressed as équivalent , time. In 1971 thèse terms were in the ratio 1:0.57, where an average speed of 40 mph was assumed. 034 1.6
In order to test the effectiveness of the linear model, the observations were grouped by the value of the linear predictor given in Table 1.2.1 into twenty-four groups, each containing between 50 and 500 observations, representing between 50,000 and 500,000 trips taken. The observed average propensity to take a trip abroad is shown plotted against the value predicted from the linear model in Figure 1.2.1. As can be seen, although the linear model explains a large proportion (98.7îO of the variance in the observed average propensities, there is a tendency to overpredict at low values of the indicator, and to underpredict at high values. Furthermore, there is an important theoretical defect of the linear model, that it can (and does) predict négative propensities.
The method used to overcome thèse defects was as follows. The linear combination of socio-economic variables derived from the régression above gives us an indication of the direction in socio-economic space in which the systematic variation of propensity to take a leisure trip -abroad is greatest. However, this variation is unlikely to be strictly linear. A function which has the correct properties to overcome the defects identified above is the logistic function. A non-linear estimation technique was used to estimate directly the three parameters of this function, using the twenty-four grouped observations. This technique gave the logistic function
1 ' propensity = - 7.259 x 21.91e + 1.173 where x is the best linear estimator, and increased the explanatory power to 99.3/5 of the variance. Given the more désirable asymptotic properties of this function, our use of it is justified. The best logistic predictor is plotted in Figure 1.2.1. The parameters estimated give an asymptotic propensity of 0.853 trips, a reasonable figure. However, we appreciate that it is important to know the reliability of such a critical parameter, and the sensitivity of our prédictions to its value. Accordingly, we tested différent models, re-estimating the other parameters, for a saturation level considerably higher (1.25), and for a function without saturation. The higher asymptote gives a fit nearly as good (0.9$ unexplained variance, as opposed to 0.7#), but the non-saturating function gives a very much worse fit (2.1$ unexplained variance). We are therefore confident that a model which tends to an asymptote is the best représentation of the data, but the sensitivity of any prédictions made to the value of this asypmtote ought to be considered. 035 Ll Figure 1-2-1 LineQr and logistic models of holiday demand (data from 1971 Household Survey)
! i 1-0 — y
:- - - asymptote I i 0-8 logistic modeL^ /
0-6 V / linear model Observed propensity of group 0-4
0-2
00 r y** ii i i i -0-2 0 0-2 0-4 0-6 08 10
Best linear estimator 036
1.2.3 Geographical Distribution of UK Lei3ure Trip Demand
A problem arise3 from the use of the 1971 data base in forecasting future geographical distribution of the demand for UK leisure trips abroad. The régions recorded on the 1971 survey file correspond to administrative boundaries in use in 1971. However, in 1974 and 1975, there was a gênerai reorganisation of administrative boundaries in the UK, with the resuit that the zones and régions recorded in the survey could riot be reconciled with those for which data are now available, both for validation and for forecasting exogenous variables. We therefore decided to forecast the régional distribution of UK leisure trip demand on a more aggregate basis, using data on average socio-economic characteristics for counties which were readily available from published statistics. The counties of Great Britain were aggregated into twenty-five zones, using uniformity of route choice variables for the Channel crossing as the principal criterion.
The zones chosen are shown in Figure 1.2.2. A model of propensity of to make a leisure trip abroad was estimated using the IPS 1977 statistics with socio-economic characteristics for counties obtained from the CSO publication, Régional Statistics. Variables found to be significant were an income variable, a car ownership variable, and a geographical variable. The final form selected was as follows:-
0.5736 1.396 0.9135 P - 1.136 C I D where P was the propensity to take a trip abroad (in return trips per person per year); C was the car ownership for the zone (in cars per person); ï was a measure of income (the only readily available measure being average weekly earnings); D was a measure of accessibility (the best found being 160 miles or distance from Dover, whichever was the smaller).
This form explained 94% of the variance of the logs of P, with 25 observations. Results of this régression are shown on Figure 1.2.3.
The total demand for leisure trips abroad will be forecast using the model described in the preceding section. The model in the présent section will then be used to détermine the allocation of this demand to origin zones; the propensities predicted will be normalised to the total demand given in the first stage. 4 u J l j ^ J
Figure 1.2.2
to
67 68 69
70
72
76
t 52 | 73 O Hgure 1.2.2 Zcning System for route choice model il 038 [] Figure 12 3 Propensity to take leisure trips abroad by U.K. région u (data from IPS tabulations, 1977)
x Surrey 1,1 400
300 —
Observed X / propensity / X X JK X 200 ~ x S
* y y x/
x s 100 5X /\ /* X / * X
0 1 1 1 0 100 200 300 Predicted propensity to take leisure trips abroad (return trips per thousand population) 033 1.8
1.2.4 The Model of Trip Type
Because of apparently large changes in the balance between différent types of trip between the 1971 survey and 1977 IPS information, and a complète lack of time-series information, we decided to explain the split between the three types of trip - car, independent non-car, and package - using the 1977 IPS data, and using as explanatory variables zonal average socio-economic characteristics. The modelling was carried out in two stages; the first stage predicts the proportions of leisure trips from a given U.K. zone with and without a car, and in the second stage, the non-car trips are split into independent and package. In the first stage the explanatory variable was car ownership, and in the second, an index of accessibility to the continent. The régressions are summarised in Table 1.2.3» The proportions of the variance explained are 58.1$ and 71$ respectively which suggest that much of the measurable effect is being accounted for. The performance of the models is illustrated in Figures 1.2.4, 1.2.5 and 1.2.6.
At the first stage, it is found that, not surprisingly, the proportion of car trips rises with increasing car ownership. We found no systematic geographical variation in this proportion. For the split of non-car trips into independent and package, the important factor appears to be geographical. We examined the effect of distance from Dover, and found that at distances from Dover greater than 160 miles, non-car trips are split approximately 40$ independent, 60$ package. For zones nearer Dover, the proportion of independent trips increases. Other geographical variables were also examined, and found to give a less satisfactory explanation. It should be noted that for most of the country, distance from Dover is highly correlated with distance from Heathrow and Gatwick. The effect is probably attributable to an increasing number of shorter trips, which are mostly in the independent category.
Table 1.2.3: Fractions of Trips Abroad by Type
Car Accessibility % of Variance Stage Holiday Type Intercept Ownership Index Explained
First Car -0.03701 0.61
Non-Car 1.03701 -0.61 ^\\ 58.1
Second Independent 0.7498 -0.00220 70.1 Package 0.2502 O.OO22O \\\ 040
Figure 1-2-t Leisure trips abroad by car as a percentage of ali leisure trips abroad
20 Devon & Cornwall,
15
Wales X Observed X car fraction Beds x x x 0/ /o Norfolk
5
0 5 10 15 20
Predicted car fraction % 041
Figure 1-2-5 Independent non-car leisure trips abroad as a percentage of total non-car leisure trips abroad.
70 _
x /
X / Oxford x 60 Hantsx X X x y^x Observed X independent 50 X fraction X S x A X yT Essexx X 40 y s* S Sx * /York * 30 / & Humber II 1 30 50 60 70
Predicted independent fraction (%) 042
Figure 1-2-6 Package leisure trips abroad as a percentage of total non-car trips abroad
York & Humber. / :xxx \ 60 /x X Observed Essex x S X package x yf 50 X fraction / X / X x/ X X x ^ xHants CO . / x Oxford S X y x 30 / i 30 50 60 70
Predicted package fraction (%) 043 1.9
Othér possible explanatory variables examined included income, and différent measures of proximity to urban centres. No useful relationship could be found with thera.
The model implies that independent and package holidays will reraain in a ratio for each U.K. zone which does not change. This seems quite reasonable given the trend in independent and package holidays since 1972. It also implies that growth in the proportion of car holidays will level off at the same time as car ownership does.
1.2.5 The Model of Destination Choice
We examined the changes in the proportions of travellers in each category choosing a given destination country over the period 1972 - 1 977. A. number of variables were found to be useful in explaining variations between countries in attractiveness to UK leisure travellers. However, thèse variables represented unchanging factors such as présence of coastlines, and as such were not useful for predicting any changes in the patterns of destination choice which might occur, or for explaining those which had occurred. It was apparent that an important factor in determining thèse changes was the change in relative cost of living. Figures 1.2.7 and 1.2.8 show how destination choice has varied for independent and package holidays, and Figure 1.2.9 shows the variation in a relative cost-of-living variable. This is defined as the ratio of consumer price indices in the destination country and the UK, expressed relative to 1970 and corrected for exchange rate changes since 1970. Since there are only six observations per destination, it was not possible to examine each country séparately. Furthermore, for most countries, (France, Belgium, West Germany, the Netherlands) the variation in both the price variable and the proportions were sraall. For Switzerland and Austria, and to a lesser. extent Spain, an increase in cost had been accompanied by a drop in share, and for the US and Canada, a decrease in cost had been accompanied by an increase in share. However, Italy is a major exception to the expected pattern. While the cost of staying in Italy has declined since 1972, the proportion of travellers going to Italy has declined steadily also. This trend is part of a décline which has continued since the early 1960s.
Since, as will be seen, the scénarios agreed by-the Commission for which prédictions are needed postulate no change in this relative cost-of-living variable, we did not consider it important to examine the influence of this variable in détail. 044 Figure 127 Destination choice of UK leisure travellers (independent) 34 •France
32
30 ^^ / V
10 Share of traffic West Germany C/.J """'••Spain 8
6 —USA ;Htaly Netherlands Belgium Canada
•—Switzerland
•—Austria
1972 1973 197C 1975 1976 1977 Key: Rising shares (France, USA, Canada) Constant shares (Germany, Netherlands, Belgium) Falling shares (Spain, Italy, Switzerland, Austria) 045 Figure 12 8 Destination choice of UK leisure travellers (package)
50
Spain u
12
• Italy Share of traffic •France 8
•-Greece
' Ausiria
2 Belgium Netherlands •Switzerland West Germany
1972 1973 1974 1975 1976 1977 Key: — Rising shares (France, Greece) — Constant shares (Belgium, Netherlands) ^ •— Falling shares (Spain, Italy, Austria, Switzerland, Germany) 046 Figure 1-2 9 Index of real cost of staying abroad for UK résidents (1970 = 100)
SWITZ.
NETH. AUSTRIA
BELGIUM
•SPAIN GERMANY
FRANCE Weighted averoge
JÏALY "GREECE 04? 1.10
A question which deserves examination is the influence of travel cost différences on destination choice. There are two difficulties in exaraining this. The first is that of selecting a suitable independent variable which will accurately reflect changes in the cost of travel. The second is that the effect of variation in-such a variable appears to be very small compared with the variations arising from unraeasured factors. We were unable to detect any such effect in the séries examined.
An alternative approach is to estimate the parameters of an explicit distribution model. Such a model explains the trips made frora a given production zone (UK zone) to a given attraction zone (continental country) in terras of the total trips produced by the production zone, the total trips attracted by the attracting zone, and the disutility of or impédance to travel between the zones. Such a raodel has the added advantage of explaining the différent patterns of destination choice observed for différent UK zones. Exaraination of the 1977 IPS data for leisure car trips indicated that such a model, with an implicit elasticity to travel impédance of about -3, would be a good explanation of observod différences between UK zones in patterns of destination choice. However, we were unable to find such a pattern for non-car trips, and given that our data was only disaggregated to continental country level, and also the difficulty of combining this approach with our "top-down" demand models, it was decided to neglect the effect of changes in travel cost on destination choice.
To apply a route choice model sensibly, it is necessary to hâve some knowledge of destination choice at a more detailed level than country. There appears to be no source of information on this more récent than the IPS trailer survey undertaken for the previous study in 1971. Unless further information becoraes available, the split of traffic in a given category going to a given country between its various régions must be assumed to be that observed in 1971. The zones used are shown in Figure 1.2.2.
1.2.6 The Model of Route Choice
The analysis of route choice behaviour in the context of Channel crossing differs from most studies in the large number of possible choices open to any individuel traveller. Travellers from a given UK zone to a given continental zone are observed to use many of the available routes, and it is quite possible for a single individual to choose différent routes for his outward and return journeys. Accordingly, it was necessary to use a non- deterministic raodel, in which the route chosen by a given traveller was not siraply a function of his measured characteristics; instead, travellers with 048 LU
given characteristics are assigned probabilities of choosing a variety of routes.
For the purpose of route choice, we divided leisure travellers into three catégories:
- car travellers; - independent, non-car travellers; package travellers.
The models used for thèse three catégories hâve the same basic structure, as does that for freight route choice.
1.2.7 The Route Choice Model for Car Travellers
Networks of transport links were constructed, which enabled us to calculate, for the traveller from UK zone i to Continental zone j, the characteristics of ail the possible routes. Thèse characteristics include:
- mean UK road distance; - mean Continental road distance; - mean time taken for crossing; • - mean crossing fare. We assume that any individual wishing to drive to the Continent chooses a route by rainimising some function of his crossing characteristics. However, his individual values will not be the same as the mean values calculated from the network. Furthermore, the form of each individual's impédance function may be différent. We simulate this by postulating a form of an impédance function, into which we insert the mean values calculated as above. Such a function is:
I. . = a D? + a DC. + a T. + a .F, ij u îp c qj t k f k where Ijj is the mean impédance for travellers from zone i to zone j using crossing k; - Duip is the UK road distance from zone i to the port p at the UK end of crossing k; - Dcqj is the Continental road distance from the port q at the Continental end of crossing k; - Tjç is the time required for crossingk; - Fjç is a typical fare for.crossing k;
- au is the impédance per unit of UK road distance;
- ac is the impédance per unit of Continental road distance; - a-j; is the impédance per unit time; - af is the impédance per unit of typical fare. 04 9 1.12
u c Values of D , D , Tjj and Pk may be obtained from the network. The fare we hâve chosen is that for two adults and two children in a 14-foot car, on a summer weekday. The other variables, au, ac, a^ and af are parameters to be detennined from the observations. Clearly, they are not ail four independent. It is convenient to set the average af to unity, so that the impédance is expressed in money units.
Variation in the routes chosen by individuels with common measured characteristics can arise from many sources. The parameters au and ac represent the sum of two effects. They take account of the running costs of the car (which will, of course, vary with the size of the car) and of the perceived disutility of time spent travelling in the car, which will vary with the number in the party, and, subjectively, with their perceptions of time. Similarly, a^ will vary with the size of the party and'with their perceptions. of time spent on a ferry (for example, a^ will be high for an individuel liable to sea-sickness, but low for one to whom it is a pleasure to travel by sea). The value of af will be low for an individual in a sraall car, or a shareholder claiming a concessionary fare, but high for a large party touring in a caravan. As well as parameter values, the values of the observable variables may differ from the mean; an individual in a zone may live nearer to further away from the port than the zone centroid.
It is clearly not practical to model ail the sources of uncertainty described above individually. We hâve therefore chosen to simulate the uncertainty arising from ail possible factors by assuming that individuals' impédances for ail the routes are distributed about the means according to prescribed distributions. Each individual takes the route with the lowest impédance. For the population, the probability of taking a given route can be calculated. Détails of the calculation, and other material pertinent to route choice, can be found in Appendix B.
Two crucial questions now arise:
- what is the structure of the distribution? - what is the form of the impédance function? 050 n
We first consider the distribution of déviations frora the raean. For reasons . ] described in Appendix B, we hâve assumed that the distribution is normal, with ^ standard dérivation proportional to the mean. The approach suggested, if p applied unmodified in the context of Channel crossing, can lead to soraewhat . implausible results, because the analysis assumes that the déviations from the mean for différent routes are independently distributed. However, it is likely j that the déviations from the mean for a pair of crossings like Dover-Calais and Dover-Boulogne (and similarly, for the new link) are highly correlated.
] Our approach to this problem has been to split the route choice into more than one stage. At each stage we try to represent choices in which each j one genuinely is irrelevant to choices between the others. The stages chosen for our car model are shown diagrammatically in Figure 1.2.10. I \ The formulation chosen does not lend itself to formai statistical estimation. The procédure used was to examine différent forms of impédance ! function, and différent parameter values, and to choose, by trial and error, the function which minimises the number of trips wrongly allocated to u; crossings. The functional form arrived at was an additive generalised cost î impédance function: j I = 0.14 Du + 0.1 Dc + F + 1.35T
\ where Du was the road distance in the UK in miles; - D was the road distance on the Continent in miles; i - F was the peak weekday fare for two adults and two children in a 14-foot car, in 1979 pounds sterling; - T was the crossing time, up to a maximum of 12 hours. Forms with the same coefficient for IÎK and Continental distance were tried, but ' found unsustainable. Figure 1-2-10 Route choice stages for car travellers
1 1 1 1 1 Long sea routes Harwich and Routes Sheerness- East Kent ports Newhaven- Hampshire ports Weymouth- Plymouth- CORRIDORS to Scandinavie, from Hull Felixstcv/e - Vlissingen Germany&Spain Belgium&Neth. Belgium & France Dieppe - France Cherbourg Roscoff
SUCCESSIVE STAGES French ports Belgian ports
1 1 SUCCESSIVE STAGES Calais (ship) Boulogne (ship) Hovercraft 052 1.14
The implications of the parameter'values found are interesting, especially with regard to people's valuations or perceptions of time. The values of ac and au subsume the perceived costs of monitoring and the disutility of time on the road. If a plausible average value of 4p per mile for the running costs is assumed, this leaves disutilities of time in the car of 1Op per mile in the UK and 6p per mile on the Continent; that is, £4 per hour in the UK and £2.40 on the Continent, for the average car-load at an average speed of 40 mph. This différence perception of driving in the UK and the Continent accords with the findings of previous studies. With an average occupancy of just over three persons, the "values of time" found are higher than those usual in studies of ahorter journeys, which is not at ail surprising. A much lower disutility of time on the ferry is found, which again is intuitively reasonable, since on average travellers seem to enjoy the ferry trip more than they do driving.
The impédance function chosen to fit the split between corridors was also found to model adequately the second stage (the split between French and Belgian ports out of East Kent ports) and the third stage (the split between Boulogne and Calais out of Dover), provided that a factor allowing for the différent frequencies is incorporated. Détails of the fit are given in Appendix B. In addition, it was found necessary to allow for the fact that Calais had better road connections than Boulogne.
As expected, the coefficients of variation found in the three stages are différent. For the choice of corridor, the best distribution was normal with a standard déviation 0.026 of the mean. For the choice between French and Belgian ports, a value of 0.016 was found. This lower value reflects the fact that routes via thèse crossings hâve a great deal in common, including the journey in the UK, and a common fare structure. For the split between Calais and Boulogne, the best value was 0.031» which, interestingly, is greater than that for the Belgian/French split. This may reflect the lack of information people hâve on the relative advantages of Calais and Boulogne, whereas those of the Belgian ports vis-a-vis the French ports may seem more obvious.
1.2.8 The Choice Between Ship and Hovercraft
Unfortunately, the only detailed data we possess on the patterns of use of the hovercraft and ships out of East Kent ports are for 1977, a year in which the Seaspeed hovercraft services were severely curtailed. No consistent pattern of choice between hovercraft and ship could be discerned in the data. f] 053
[1 »-J In the choice of hovercraft or ahip were determined by the same impédance , function we hâve used hitherto, the hovercraft could potentially attract a [_J large share - perhaps 60$ - of the short-sea market. This could indeed be the case, and the factor constraining the hovercraft share might simply be lack of ' j capacity. On the other hand, the disutility of time spent on a hovercraft may well be différent frora that spent on a ship for reasons of discomfort, or J j unreliability. The data we hâve gave no way of determining this. Since 1977, I j . the hovercraft operators hâve started to charge a small preraiura (about 3% on f} the standard car-load) for travellers in the peak period. In addition, the '• - hovercraft are at présent running at a loss, or, at best, making a very srnall return on fheir capital. Thus it is likely that they will in future charge ) perhaps 5-10$ more than the ferries, and take a correspondingly smaller market share. ! ) 1.2.9 The Route Choice Model for Independent Non-Car Travellers i") For this category of travellers, we hâve adopted a model of the same i i form as that just described for car travellers. Data were again assembled and r-, stored as a transport network, détails of which are given in Appendix B. i i ! - Scheduled rail and air services were used, with some non-scheduled ,---. air services where large flows were observed on a route where no scheduled i , service existed, for example between Manchester and Spain. Explicit allowance was made for waiting and transfer time-. i i Crossings were grouped into corridors as for car travellers, with the addition of corridors for air travellers (one for each UK airport considered, / save the London airports which were treated as a single corridor). Factors were incorporated to allow for the différent proportions of charter flights available from différent airports.
The impédance function finally used was:- I « T + £ V where T is the through time including rail time, air or ferry time, time for access to railhead or airport, and waiting time; C is the through cost, including rail, air and ferry fares; and V is a parameter, the "value of time". 054
In principle, this impédance function could be formulated to allow V to be différent for différent types of time, e.g. time in train, time in ship, time in flight, waiting time. However, this complication was not found helpful.
The parameters of the model were V and the coefficient of variance. The standard déviations of the impédance distributions were assumed proportional to the means throughout. The best values of thèse parameters were found by searching for those which minimise the trips recorded in the 1977 IPS tabulation wrongly allocated to a corridor.
A modification was necessary as it became apparent that, for example, the disutility of a fifteen-hour train journey was much more than three times that of a five-hour journey. Thus values of time which gave an acceptable rail-air split for short journeys, e.g. to Belgium, the Netherlands and Northern France, gave implausibily high proportions on rail to Austria, Italy and Spain. After investigation of a number of possibilities to find the best explanation, this was overcome by increasing the disutility per unit time by 50$ for journeys over eight hours. This "factor reflects in part the cost of sleeper accommodation, etc., and in part the sheer discomfort of long train journeys. It should also be borne in mind that the journeys considered are usually part of a holiday which has a fixed total time allocated to it. This factor will certainly increase the travellers' résistance to spending a very long time on a train.
The velue of the parameters which gave the best fit were a value of time of £1.50 per hour for an individual, and a standard déviation of 1% of the raean impédance. The value of time implied is much higher than those found in many other studies, and is higher than that found for car travellers abroad. However, this is not surprising, given the constraint on the total time which usually applies to holidays; thus time spent travelling is holiday lost. In addition, it should be recalled that this value of time is expressed relative to full scheduled fare, and that in practice a large proportion of travellers will avail themselves of discounts and cheap fares. Thus the "real" value of time, that is, that which should be used for évaluation of benefits arising from réductions in travel time for this category of traveller will be proportionately lower.
1.2.10 The Route Choice Model for Package Travellers
The behaviour of this category of traveller is more difficult to model than that of the independent traveller because, by définition, the information on fare paid is not available. The characteristics of the package traveller are that: 055 1.17
- he tends to travel more by air; - he tends to use régional airports rather than Heathrow; and - when travelling by sea, he tends to use the longer sea crossings to a greater extend than the independent traveller, at the expense of the French and Belgian straits routes.
Thèse characteristics hâve interesting implications. If package travellers were asauraed to pay the same fares as independent travellers they would appear to hâve a higher value of time parameter. A more likely explanation is that there is, in fact, a substantial réduction in the fares actually or iraplicitly paid, which would hâve the same effect. Another factor working in the same direction is the fact that a much larger proportion of package travellers use coach, a slower, cheaper mode than rail. Thèse implicit réductions in fare would tend to shift travellers from sea to air, and, for those using sea crossings, away from the cheaper crossings (the Prench and Belgian straits).
We hâve chosen te model this behaviour using the same network and the same functional form as for the independent non-car travellers, but with the relative importance of fare reduced. The best parameter values found are £2.10 per hour for the value of time, and 0.1 for the coefficient of variation; again, of course, this is relative to scheduled fares, and the "real" value of time required for evaluating benefits will be considerably lower. 056 1.18
1.3 The Models; The SETEC Model for Continental Leisure Travel to the UK
The models of demand and of route choice for Continental leisure travellers are described in the Phase 1 and Phase 2 reports by SETEC Economie. 057
1.4 The Models ; Business Travel Across the Channel
1.4.1 The Model of UK Business Travel Demand
The data from which this raodel was derived came from published IPS statistics. It should be noted that IPS published statistics for business travel include lorry drivers, and exclude certain purposes which we would classify as "business" (e.g. travel to take up employment). However, we hâve assumed that thèse can be accounted for as a fixed proportion of the published figures for each destination country.
The model used explained the number of UK-origin business trips to a given country in terms of the GDP of the destination country and the value of UK exports to that country.
0.2127 0.4780 where Tj is the number of business trips (thousands of return trips) made by UK résidents to country j; A-j is a country-specific attraction factor; Gj is the value of the gross domestic product of country j, expressed in billions of 1975 US$; Xj is the value of exports from the UK to country J, expressed in billions of 1975 US$.
The model was estimated using data from 1972 to 1977, for the nine countries which contributed most to those business trips likely to use a fixed crossing: France, Belgium and Luxembourg, Netherlands, West Germany, Italy, Denmark, Spain, Switzerland and Austria.
Earlier models tested included the value of the UK GDP, and the value of UK imports from the country, as explanatory variables; thèse were not found to hâve any significant explanatory power. A model without the country-specific factor Aj was found which explained 90$ of the variance of the logarithms of Tj. However, we noticed that the average of the residuals for this model showed a significant between-country variance. The country- specific factors were therefore introduced, and improved the explanation to 99$ of the variance. The values of Aj found are listed in Table 1.4.1, and the predicted and observed numbers of trips are compared in Figures 1.4.1 and 1.4.2 for the seven moat important destinations. 058 Figure K-1 UK business trips abroad, predicted and observed
36(1
France 3U)_
320 _
300 _
West 280 Germany Business trips (thousands) 260
240 1
220 _
200
180 1972 197/, 1975 1976 1977 Key: Year Observed Predicted 059 Figure 1-4-2 UK business trips abroad, predicted and observed
Netherlands 200 L
180 L
Belgium & Luxembourg 1601
uoL Business trips (thousands) 120
100 £ Italy
Switzerland 80 L
Spain 60 L
1972 1973 1974 1975 1976 1977 Key: Year Observed Predicted 060 1.20
To correct thèse figures for the inclusion of lorry-drivers, and the exclusion of emigrants etc., from the IPS published figures, we used the proportions falling into thèse catégories for each country in 1977. They are shown, together with the modified attraction factors, in Table 1.4.2.
Table 1 .4.1 : Country-Specific Business Travel Attraction Factors
Attraction factor Aj Destination (thousands of return trips)
France 49.7 Belgium & Luxembourg 36.3 Netherlands 39.9 West Gerraany 36.3 Italy 24.3 Denmark 12.7 Spain 26.9 Switzerland 22.4 Austria 12.4
Table 1.4.2: Business Travellers in 1977: IPS and our Définitions
Business trips Business trips Modified Destination (IPS) (our définition) attraction (thousands) (thousands) factor
France 349 306 43.6 Belgium & Luxembourg 166 154 33.6 Netherlands 193 179 37.0 West Germany 285 280 35.7 Italy 93 83 21.7 Denmark 33 33 12.7 Spain 61 60 26.5 Switzerland 61 60 22.0 Austria 20 20 12.4
Total 1261 1175 ! ! ! i 061
The attraction factors obtained vary over a wide range (12.4 to ' ' 43*6). It has been found that much of this variation can be explained by the relative proximity of the country to the UK. The variable we hâve used to f -• measure this is the air fare (1977 scheduled air fare) between Heathrow and an ' i i j appropriate commercial centre. The centres chosen were Paris, Brussels,
Amsterdam, Koln, Milan, Copenhagen, Madrid, Zurich and Wien.
• ' The most appropriate simple model form appears to be: A. = a.e J J
where F-j i3 the air fare.
This has better asymptotic properties than the alternative linear or log-log forms. The Aj values found are well represented by this relationship, save those for Spain and Italy. ¥e therefore take into account variations in air transport costs in forecasting the A values, by using:
A.(t) , A. *rfe hâve assumed that the residual variation found in the Aj values is due to some factor which will remain, on average, constant. IPS figures for 1 977 also give us the breakdown of business travel by UK origin. Thèse figures are shown in Table 1.4.3. In the absence of further data, we assume that thèse proportions reraain fixed in the future. 0C2 1.22 Table 1.4.3: UK Business Trips by Origin, 1977 Business trips Origin (thousands) * of UK Total Scotland 41 3.5* North 21 1.8* Nofth-west, NI, IOM 97 8.3* Yorks & Humberside 54 4.6* East Midlands 58 5.0* West Midlands 92 7.9* Norfolk 7 0.6* Carabs 17 1.5* Beds 15 1.3* Bucks 35 3.0* Herts 46 3.9* Suffolk 13 1.1* Essex 57 4.9* Oxon 15 1.3* Berks 37 3.2* London 247 21.1* Surrey 103 8.8* Kent 68 5.8* East Sussex 17 1.5* West Sussex 26 2.2* Hants & IOW 37 3.2* Dorset 8 0.7* De von & Cornwall 5 0.4* Rest of South west 36 3.1* Wales 17 1.5* Total 1169 100.0* Notes: (a) Business trips are defined as return trips to the nine destinations in Tables 1.4.1 and 1.4.2; they correspond to our définition; (b) The total does not correspond to that in Table 1.4.2 becasue the latter includes 2,500 trips originating in the Channel Islands, which will go by air, and 4,500 "dont't know origin" trips. 0C3 1.23 J 1.5 The Models ; Non-European Travel Across the Channel 1.5*1 The Model of Non-European Travel Demand An attempt to construct a model of the demand for travel between the ; U.K. and continental Europe by non-Europeans is beset by considérable problems. First, and most important, there appears to be no source of data on the level î of demand. Secondly, what data there are on the movements of non-Europeans within Europe, are only available on a very patchy basis; each data source ; uses a différent aggregation of countries, and différent définitions. We hâve relied on four data sources:- - IPS published statistics; - OECD travel and tourism statistics; - USA and Australian data on departure for Europe; and - surveys undertaken for previous Channel Tunnel studies. IPS published statistics give the number of non-Europeans entering Britain, but do not break thèse numbers down into those airiving from Europe and those "••; arriving from outside Europe. —' The classification of non-Europeans is not consistent from year to '• year, but long séries are available from the USA, Canada, Australia and New Zealand together, and South Africa. More recently visits from Japan and Latin America hâve been given, and very recently, séries for the Middle East, North Africa, and Other Africa. The séries are shown in Figure 1.5*1* The USA provides the major source of non-European visitors (41$) but totals dropped after a rapid rise in the late 196O's. Visitors from Canada and Australia al3O rosé rapidly over the same period, and the Australasians were apparently not affected by the factors which caused the drop in North American visitors. The rise in the number of visitors from the Middle East has been exceedingly rapid since 1975, when it was first published. Of the over four million non-Europeans visiting the U.K., it is difficult to détermine what proportion will make the crossing to Continental Europe. The OECD publication, Travel and Tourism Statistics, publishes some figures collected by the member nations. Nations vary in how they collect tourism data; for example, most give numbers of foreign guests registering in hôtels etc. Some, however, for example the U.K., count frontier crossings and others, for example Belgium, give only the number of nights spent in accommodation. Furthermore, the nations differ in the extent to which they aggregate non-European visitors; France and Spain give separate détails of a large number of origin countries, while Belgium only uses four catégories. 0G4 J Figurc: e 1-5-1 Non-European visitors to BritQin (source: IPS) L LJ 1800_ 1600. LJ 1400. 1200 _ Thousands of visits 1000. Canada ^Middle Easi \Australia & New Zealand Resf of World Other Africa ,.*-Latin America South Africa vJapan ^North Africa 196A 1965 1966 1967 1968 1969 1970 1971 1972 1973 TO 1975 1976 1977 0G5 1.24 J By making plausible assumptions where data are inadéquate, and •i neglecting those viaitors who do not stay in tourist accommodation, it is J possible to estimate the numbers of frontier crossings by non-Europeans for 1976, the latest year for which figures are available. They are given in Table j 1.5.1. Table 1»5«1: Estimated Frontier Crossings by non-Europeans, 1976 J Frontier Crossing (thousands) 1 USA 10749 ' Canada 2150 ^ Australia & N.Z. 1264 J . Japan 1570 Latin America 1715 | Rest of World 4530 Source: OECD Travel and Toorism Statistics, 1977. i Countries included are: Austria, Belgium, Denmark, France, West Germany, Greece, Italy, Netherlands, Portugal, Spain, Switzerland, U.K., Yugoslavia. ' For the previous Channel Tunnel Studies, information was provided by • ••• certain non-European nations on the number of their nationals visiting Europe. For .this study, this information was only available from the USA and Australia, and only the USA could provide a long time séries. This time séries was used to test a modification of the raodel used in the previous study: 0.51 0.97 -0.90 T = T P I C o where T is the number of trips from an overseas région to Europe; P is the population of the région; I is its GDP per head; and C is the cost of a European holiday from that région. 0G6 1.25 The performance of this model can be tested as the US Department of Commerce supplies data on the séries of costs as well as on trips. The performance is shown by the dashed line in Figure 1.5.2. An even better agreeraent it obtained by using the travel cost for C; this is shown as the dotted line. This is also more plausible behaviourally, since the main déterminant of holiday cost is length of stay, which the traveller is himself largely free to choose. The performance of this raodelmake s it quite acceptable as a forecasting method. The figures for visitors to Europe from the USA and Australia, when corabined with those for the number of frontier crossings, enable us to infer an average number of countries visited. In 1976 this was 3*3 for USA résidents, and 3.9 for Australians. This compares with the 1971 figures of 3.8 for USA and 4.0 for Australia and New Zealand taken together. However, thèse figures included an "estimate for countries not recorded". Divergences are not too great for us to accept the 1971 estimâtes for countries per trip for other parts of the world. Our estimâtes of numbers of non-Europeans visiting Europe are given in Table 1.5.2. Table 1.5.2: Estimated Trips to Europe by Non-Europeans, 1976 - Countries per trip Estimated. trips (thousands) USA 3.3 3257 Canada 2.9 741 Australia & N.Z. . 3.9 414 Japan 5.5 285 Latin America 6.0 286 Rest of World 6.0 755 5738 The final step required is to calculate how many trips from continental Europe to Britain are generated by thèse travellers. The method used is to assume that each time a non-European traveller enters a continental country, there is a fixed probability that either his next destination will be Britain, or his immédiate origin was Britain. Thèse probabilities are derived frora the 1971 surveys. 0G7 Figure 1*5*2 US travellers to Europe (source: US Dept of Commerce) 6 5 Millions of trips 3 2 I I I I I I I I I 1965 1966 1967 1968 1969 1970 1971 1972 1973 197/, 1975. 1976 1977 YEAR Key: Observed Predicted (model with holiday cost) • • Predicted (model with travel cost) 0G8 D 1.26 _!• The model proposed can be summarised• as follows: M 1. Trips from a non-European origin to Europe dépend upon the k-j population of the origin région, its GDP per head, and the — cost of travel to Europe. (j 2. Thèse travellers are assumed to visit European countries in the same pattèrns as they did in 1976; thèse patterns are ' J summarised in Table 1.5.3. 3« Each arrivai in a given continental country is assumed to j I generate a fixed probability of a trip to the U.K.; thèse ~ probabilities are given in Table 1.5.3. j | Results of applying this procédure in 1976 are shown in Table 1.5.3. They are consistent with (i.e. always considerably less than twice) the numbers i , of return trips to the U.K., recorded by the IPS. In addition, a realistic maximum value on the number of sea crossings can be set by considering crossing ;-] to or from countries within a realistic day's journey of e, port Connecting with LJ Britain. Thèse are France, Belgium, the Netherlands, and West Germany and Scandinavia; this suggests that the number of single sea crossings raade by '•[_j non-Europeans cannot be greater than 1,650,000. This again is consistent with IPS figures, which give 3,661,000 visits to Britain made by non-Europeans of ! \ which 677,000 are by sea, which implies at most, 1,345,000 single crossings (probably less because of direct sea passengers). 0G9 1.27 1.6 The Modela ; Freight 1.6.1 The SETEC Model for Demand for Freight This is described in the reports of Phase 1 and Phase 2 by SETEC Economie. 1.6.2 Route Choice for Freight Traffic The problem of forecasting the patterns of travel of freight traffic in futures with and without a new fixed Channel link is in many ways similar to that for passengers. There is, however, an additional difficulty in that there is no single, consistent, source of data on freight movements at the level required, corresponding to IPS for passengers. An additional first stage in the dérivation of a model of freight route choice was therefore the production of a breakdown of the freight movement patterns in the base year, 1977, consistent as far as was possible with the sources of data, with which the route choice model1s performance could be compared. The development of a route choice model for freight movements between the UK and Continental Europe required the following opérations:- (a) analysis of the patterns of cross-Channel freight movement in the base year, 1977; (b) examination of those characteristics of existing methods of transport (principally cost and time) which were thought to détermine route choice; and (c) dérivation of a structure, and estimation of the parameters, of a model which explained adequately the patterns found in (a), and was able to represent the relevant characteristics of a new link. Exaraination of the data available suggested that three transport modes for freight should be distinguished. Thèse were:- (a) ro-ro (roll-on, roll-off) traffic, where road vehicles carrying the freight were "rolled" on and off the vessel; (b) lo-lo (lift-on, lift-off) traffic, where containers were lifted on and off the vessel, regardless of the mode used to transport the containers to the port; and (c) rail wagon traffic, where rail vehicles were rolled on and off the vessel. 070 The second category, lo-lo traffic, will sometimes loosely be referred to as "container" traffic, because of the specalised nature of the vessels required. However, it should be remembered that much of the traffic in the other two modes will be containerised. Non-unitised traffic has been regarded as not suitable for a new fixed link and therefore its route choice has not been considered. It has been necessary, therefore, to consider the proportion of future trade which will be unitised. 1.6.3 Patterna of Freight Movement in 1977 Three principal data sources were used:- (a) National Ports Council (NPC) published data; (b) unpublished, but freely available, data collected by HH Customs and Excise, which gave UK imports and exports to spécifié countries, by UK port and by commodity; and (c) a pilot survey undertaken in 1976 for the Department of Transport by the National Ports Council; the full survey took place in 1978 and 1 % of ail traffic was sampled, but results were not ready in time for use in this study. To calibrate the test freight route choice models, it was necessary to produce a breakdown of freight flows by UK zone of origin or destination, by Continental zone of origin or destination, and by mode and route of crossing. No single source of data could supply this information. It was necessary to combine information from the various sources mentioned, and to raake inferences where there was no direct source of information. The tabulations provided by HM Customs and Excise provided flows from each UK port to each Continental country, classified by commodity and by mode; they only distinguished ro-ro (road and rail) from lo-lo traffic. However, detailed information on rail wagon flows was provided by British Rail. Customs information suffered from some important drawbacks:- (a) it is measured in "net" tonnes, that is, exclusive of packaging; 071 1.29 (b) it excluded flows of motor vehicles for import and export, and other freight rolled onto vessels by vehicles belonging to the ship or the port; (c) it excluded entrepôt traffic with ultimate destination beyond our study zone, which we regarded as divertible to a fixed link; and (d) it excluded Irish traffic crossing via the UK. Adjustments to the HM Customs figures were made to allow for thèse factors, so that totals for port throughput were the same as those given in NPC published data. Détails of thèse adjustments are given in Appendix C. The breakdown of flows by UK région of origin or destination was deduced by examining the 1976 pilot survey data. Traffic travelling through a given port was split between UK économie planning régions in the same proportions as found in that survey. Where UK zones were subdivisions of économie planning régions, the flows were split in proportion to population for imports, and in proportion to employment in agriculture, mining and manufacturing for exports. The same method was used to split the flows to Continental countries into zones. Largely because of the lack of data, we decided to model route choice for freight as a single unitised commodity. For some aspects of the studies, this is a drawback, because there is a wide variability in average load by comraodity; also, freight charges are generally levied per vehicle, rather than per tonne. However, the lack of information on the UK régional distribution of origins and destinations of différent commodities precluded our going into such détail. In any case, the variation in loadability was expected to apply more or less equally to ail modes, so that the distortion introduced by aggregating ail unitised commodities should not be too large. 1.6.4 The Costs of Various Modes of Freight Transport The following sections describe the results of surveys and analysis of haulage costs we carried out. Problems in ascertaining average tariff rates were:- (a) the lack of published data; (b) the complexity of the market; 072 1.30 (c) différent methods of expressing rates, by various agencies; (d) variations in discount schemes; (e) variations with size of consignment, consignor, product types, and season; (f) variations between import and export rates; and (g) carriers1 concern over confidentiality. Pollowing sections give the estimâtes we adopted after discussions with freight forwarders and operators. The rates analysed -were principally those for export. Factors such as differential backloading between countries were known to cause différences between import and export rates. We were unable to obtain précise import rates to the UK; however, we felt able to assume that the gênerai pattern of rates was similar. The tariffs deduced were therefore used for both imports and exports. Ail rates were expressed as charges per tonne. 1.6.5 Ro-ro Haulage Charges We carried out a survey of road hauliers in the UK, asking the cpsts of haulage for a variety of routes. Analysis of the response suggested that costs were approximately linear with road distance. NPC statistics suggest an average load of 12.4 tonnes for ail ro-ro traffic. Using this figure, the following formulae were derived from the road hauliers' response:- Cost per tonne of road haulage (in 1979, but expressed in 1977 priées) Great Britain: £2.23 + £0.032 per mile Continent: £4.01 + £0.037 per mile 1.6.6 Lo-lo Container Haulage Charges For this mode, the haulage charges had to reflect the fact that thèse containers could be hauled to port by road or rail. For road haulage, information from hauliers was consistent with our formulae for ro-ro given in the preceding section. Allowances were made, after discussions with freight forwarders, for average discounts, cost of hire, and C03t of terminal cartage for the container. NPC statistics gave an average weight of 12.7 tonnes per lifted-on container. For rail haulage, an analysis was made ctf Freightliner cost3, which proved to be approximately linear with total road distance (UK plus Continental). The weighted average of. road and rail haulage to port was calculated, giving a cost per tonne of:- 073 1.31 -' £3.40 + £0.0338 per mile j|' 1.6.7 Road Vehicle Ro-ro and Lo-lo Container Sea Crossing Tariffs A survey was made of sea-crossing tariffs for thèse modes. Thèse j were converted to 1977 priées, an allowance made for average discount (10»), and expressed as a charge per tonne. Information was also obtained on ferry "1 operators1 receipts per ro-ro vehicle in 1977» Thèse were found to be J generally similar to the charges calculated above, and as they appeared to be , from a more reliable source we decided to use them. Détails of the charges J obtained are given in Appendix D. 1.6.8 Rail Haulage and Sea-Crossing Tariffs ] . Rail haulage charges were found to be complex and needed extensive i• scrutiny. Rates charged were usually found to be countr'y-specific, and varied greatly. The main sources of information were the B.R. Shipping and -, International Services Division (SISD) and rail freight tariff books, which A were analysed to develop rail haulage charges (see Appendix D). Tariff books _ were available for France, Germany, Holland, Austria, Belgium and Switzerland. j j Tariffs to other countries hâve to be worked out by SISD on an individual basis. • . . A total international rail tariff may be viewed as being made up of the following elements:- ' (a) British section charges; (b) sea ferry charges; ! (c) interraediate country section charges (if any); and (d) final Continental country section charges. : A single flat formula (fixed charge, plus total land distance times variable charge rate) was used for "Near Sea" countries such as Belgium, Holland and France. More distant countries tend to hâve more complex tariffs. They may hâve two parts like the Austrian tariff (the British and sea-crossing section and the Continental section) or three parts like the German tariff ' (British and sea, intermediate country, and German sections). For the more distant countries, charges across intervening countries are individually l . negotiated. 074 1.32 The charges are usually set by wagon type (i.e. 2-axle, more than 2-axle, etc.). Anaysis of B.R. statistics showed 13 tonnes to be the average wagon weight for U.K. international traffic (allowing for emply running). Approximately 80$ of traffic is 2-axle, and therefore was taken as the average for calculation. The charge rates were generally inclusive of sea-crossing and border charges, but they did not include terminal and delivery costs. Discounts granted vary greatly with commodity, load size, type of haul, destination country, and the gênerai state of the market. The tariffs reviewed were effective from April 1979, and are revised yearly. According to B.R., tariffs had been increased by approximately 12.5$ in 1977, and 22.0$ in 1978. The tariff rates to various points within a country were analysed, and a déduction made for:- (a) the sea-crossing charge; and (b) an average discount. This left the charge for rail haulage in the U.K. and on the Continent. Using régression, the following formula fora was fitted to ail datât- Total charge = Pixed charge + (Variable Charge per mile x Total U.K. and Continental road mileage). This yielded the results shown in Table 1.6.1, expressed in 1977 priées. Table 1.6.1 Rail Haulage Charges Fixed Charge Variable Charge Continental Country (£ per tonne) per mile (£ per tonne) France 5.54 0.0236 Belgium 7.44 0.0151 Holland 11.74 0.0151 Switzerland 25.27 0.0273 Ge rmany 17.26 0.0160 Austria 21.29 0.0094 075 1.33 I.J As can be seen in Table 1.6.1, charge rates by country vary greatly. for the other countries for which no fixed tariff rate exists, the following assumptions were made: for Portugal, Spain and Italy, the French tariff was used; for Denmark and for Eastern Europe, the Gerraan tariff. Our studies suggested that using the tariff rates of thèse countries was the raost appropriate method of representing the non-published tariff rates. 1.6.9 Summary of Haulage Costs The comparison of the haulage costs of the alternative modes is given below in Table 1.6.2. Table 1.6.2 Comparison of Haulage Costs 1977 Prices Costs per Tonne Mode Fixed Variable Cost per Mile (£) (£) Ro-ro - U.K. haulage 2.23 0.032 - Continental haulage 4.01 0.037 Container: U.K. - Continent 3.4 0.034 Rail Wagons: U.K. - France 5.54 0.0236 U.K. - Switzerland 25.57 0.0273 U.K. - Germany 17.26 0.0160 In gênerai the ro-ro mode hàs a lower fixed charge, but a higher cost per mile. The most noticeable feature is the wide variation of the rail haulage costs between countries. As noted, thèse figures are ail based on flows from the U.K. to the Continent. Référence was made to other studies and agreement for the gênerai fora of the abcve tariff formulae was found, although it was not possible to find directly comparable work to validate our average rates. 1.6.10 Transit Times for Various Modes We felt that time of transit was likely to be an important déterminant of route choice. For modes other than road-hauled ro-ro, we found there was often a significant différence between timetabled and actual times. A small but significant proportion of thèse traffics suffered long delays in transit. Thèse arose from delays in marshalling and from goods being lost in transit "for days or weeks". The delays experienced by some traffics greatly affected the mean transit times we used. 07 6 i.34 -J Information was collected frora freight forwarders on road ro-ro -, transit times from London to various Continental destinations. This included J the sea-crossing and port times. Deducting thèse, and regressing time taken against total land mileage (U.K. plus Continental) gave a resuit of:- 16.0 hours + 0.058 hours per mile The fixed élément includes ail fixed times, other than the sea crossing and | port times. The data are summarised in Appendix E. j As with ro-ro, information was collected from freight forwarders on j container transit times from U.K. origins to U.K. ports, and from Continental j ports to Continental destinations. As for the costs, a weighted average of j road haulage and rail haulage times was taken. Thèse were again found to be approximately linear with distance; the best formula obtained was:- j 24.00 hours + 0.104 hours per mile i The fixed time includes collection and .delivery times. "| The transit tiraes for the Iberian Peninsula were not consistent with i -I our gênerai linear formula. Vérification from the previous Channel Tunnel -, studies, and from other sources, confirmed that transit speeds to Iberia were J at présent slower than average. An additional fixed time for Iberia of 96 hours was therefore assumed, to allow for this. î J Information was provided by B.R. on timetabled wagon transit times to 51 Continental destinations from the U.TC. (see Appendix E). Port times and sea-crossing times were deducted from thèse to give haulage times. A linear relationship with distance was found, and while small variations in the i coefficients between différent countries were found, we decided, in view of the small number of observations, to dérive a gênerai formula. Thi3 was:- ; 29.4 hours + 0.09 hours per mile Confidential information obtained from other sources and from the , British Rail Operational Research Division, however, indicated that actual rail wagon transit times were, in practice, much longer. The U.K. - Continental transit formula finally adopted for use was:- 1 48.0 hours + 0.288 hours per mile 1.35 077 Transit times for the différent modes are summarised in Table 1.6.3. Table 1.6.3 Comparison of Transit Times Transit Times Mode Fixed Variable (hours) (hours per mile) Ro-ro 16.0 0.058 Container 24.0 0.104 Rail Wagon 48.0 0.288 1.6.11 Port and Sea-Crossing Times The total transit time will include the sea-crossing time, and port-time. Average crossing times were obtained for each sea-route from the ferry opérators, and the frequency of service from each port. In considering the frequency of service from a port it was necessary to take account of services offered by Deep Sea ships, offering Short Sea capacity. Unfortunately we were not able to gain fully comprehenaive information on thèse services, due partly to the marginal and occasional nature of some of them. However, note was taken of the more regular services offered on Deep Sea vessels, and thèse are included in Appendix E. Examination of freight route choice showed that crossing frequency appeared to be a relevant factor. We took this into account by adding an average waiting time of half the average inter-ship time. Port times we adopted included allowance for:- (a) movement in port area; (b) ail waiting periods; (c) customs clearance; • (d) loading onto vessel. Information from freight forwarders showed that port times for lo-lo container and rail wagons were significantly longer than for ro-ro, primarily due to delays associated with Customs and loading. A large proportion of this traffic passed through on tiraetabled times.. However, a small, but significant proportion of the traffic experienced through-port transit times that were much greater than the mode (i.e. most common) time. This greatly affected the average times taken at U.K. and Continental ports, as shown in Table 1.6.4. 078 1.36 Table 1.6.4 Average Time per Port Mode Total port hours Ro-ro 1.5 hrs. Container 66.0 hrs. Rail Wagon 56.0 hrs. Thèse times were doubled to give the total port times incurred on a sea crossing. It is thought that times at Continental ports were slightly quicker; but we were not able to gain précise information on this. 1.6.12 The Hodel of Route Choice for Freight The route choice model for freight movements is based on the same principles as those for passengers, described above in Section 1.2.2. The freight consignor is assumed to choose the mode and route which minimises some function of the characteristics of the journey. We again acknowledge that the measured journey characteristics and average parameters cannot take account of the diversity of criteria consignors actually use; we therefore assume that the impédances or disutilities of a journey for freight consignors with the same measured characteristics "form a distribution about a mean value. Thus freight raoveraents with given measured characteristics are assigned probabilities of using several modes or routes. The impédance function used was a generalised cost function, and was made up of two éléments:- - (a) the actual cost; and (b) the total journey time taken. The parameter governing the relative importance of thèse terms, the value of tirae, is determined by examining which value gives the beat fit to 1977 observations. The actual cost is raade up of a number of éléments:- (a) the U.K. haulage cost (including terminal costs); (b) the sea-crossing cost (including port costs); and (c) the Continental haulage costs (including terminal cost.s). Thèse were derived from the formulae described above. 079 i.37 The total journey time is made up of the following éléments:- ri _j (a) the U.K. haulage time, including terminal collection and delivery; (b) the port-associated times at the U.K. ports including Customs and waiting; i (c) waiting time at port (taken to be a function of frequency); (d) the sea-crossing times; j (e) the port associated times at the Continental port, -' including Customs and waiting; and (f) the Continental haulage time, including terminal _; collection/delivery. •-j Times were calculated for the différent modes from the formulae given above. As for the passenger models, the distribution of impédances was ;, assumed to be normal, with a standard déviation proportional to the mean. The J constant of proportionality was assumed tô be constant throughout. This .'-. constant, and the value of time, were the parameters of the model. The values i that gave the best explanation of the 1977 patterns of freight movement were 3.4 pence per hour per tonne (1979 prices), for the value of time, and 0.04- for ' the coefficient of variation of the impédance distribution. The fit of this model against observations is shown in Table 1.6.5* It can be seen that the overall totals by mode agrée fairly well with the actuals. For most routes a fairly good match between actual and predicted was also obtained. 1.6.13 The Application of the Freight Route Choice Models ' Forecasts of total trade by commodity between the U.K. and the study zone were supplied by SETEC. Thèse forecasts also .included figures for the ; growth of Irish trade with Continental Europe, and figures for entrepôt trade between the U.K. and régions outside Europe which used Continental entrepôt ports. We had therefore to dérive from thèse forecasts the flows of unitised freight between our U.K. zones and Continental zones, in both directions. SETEC's original data base was the OECD published trade statistics which, while similar to HM Customs and excise figures for most commodities, included flows of off-shore aggregates and other minerais which did not pass throught the port of the country of origin. Thèse were excluded from our considération as not suitable for diversion to a fixed link. The remaining trade was compared to our unitised trade matrix, and rates of unitisation for each commodity determined. 080 1.38 To obtain future unitised trade flows, we had to forecast how the fractions unitised would change. Studies of trends in uaitisation (détails given in Appendix F) showed that unitised fractions for Near Sea traffic in f--i foodstuffs and in manufactured goods were already very high, the increase in I ! (_•} unitisation having ceased in the 1970s. On the other hand, basic matériels [1 were still largely unitised, except for a few commodities which, unlike most in J ! this class, had a high value per unit weight. A spécial study was made of those commodities for which it has been / | suggested that the présence of a a new link would induce their change to a unitised mode. This study is summarised in Appendix F. We find that the ! ; potential for such induced unitisation is low, because of:- (a) high investment in existing facilities; and i \ (b) a low value per unit weight for such commodities, which means that they would only be attracted by S [ a much cheaper form of transport. Such commodities are typically serai-bulk or bullet originating from and i ) destined for port-side installations. Contrary to initial expectation they (_ • were found to hâve little or no potential for diversion. il 081 1.39 Table 1.6.5 Fit of Freight Route Choice Model 1977 (thousands of tonnes) Crossing Imports Exports No. Description Estimated Estimated Actual Model Actual Model 1 Hull-Haraburg 114 259 103 199 2 Hull-Rotterdam 747 638 726 473 3 Haven-Hook 1488 1191 899 775 4 Haven-Zeebrugge, Dunkirk . 825 831 730 618 5 Ha rwich-Hamburg 343 416 169 267 6 Belgian Straits 1090 1388 835 905 7 French Straits 1458 1648 1119 1004 8 Newhaven-Dieppe 382 632 205 358 9 Soton-Le Havre, Cherbourg 848 1070 558 705 Total Ro-Ro 7613 8072 5485 5304 10 Hull-Hamburg 99 106 102 62 11 Hull-Rotterdam 268 298 190 292 12 Haven-Zeebrugge, Dunkirk 898 797 954 791 13 Haven-Rotterdam 775 420 593 444 14 Ha rwi c h-Ha mburg 280 386 187 227 15 Tilbury-Rotterdam 455 466 407 464 16 Soton-Cherbourg 88 44 66 320 Total Container 2863 2517 2499 2600 17 Dover-Dunkirk 400 356 118 235 18 Harwich-Zeebrugge 191 152 139 113 19 Harwich-Dunkirk 30 0 14 0 Total Rail Wagon 621 508 271 348 Total 11097 11097 8254 8254 n 082 Having determined the forecast unitised flows, we split thera by UK i .' zone and by zone within Continental country in the sarae proportions as in the *-- base year. The resulting flows are the inputs to the route choice model. 1 The route choice model also requires forecasts of any changes in times and costs of the existing modes, and the characteristics of the link. ( ; The changes in costs of existing modes are given by the agreed price i i (J scénarios, set out below in Section 2.2.3» The charges made by the link are determined in relation to the charges on the French Straits shipping routes; ( i / | however, we allow for compétitive changes in charges both there and on other crossings. Any changes in frequency occuring in response to the présence of j ,' the link must be incorporated. For links which carry ro-ro vehicles, that is, the double-track tunnel and the road bridge, an allowance must also be made I ' for the saving of driver time which results from the use of a faster i i crossing. ! ,' The model is then used to predict, for each zone-zone pair, the tonnages by each crossing, with and without the various links, and other • ' usefui information like the generalised cost per tonne, and the land raileage, by mode. c •• 083 IJ 0 i-j 0 j '"i I THE BASE FORECASTS l 084 2.1 2.1 Préface 2.1.1 Division of Respon3ibility The division of responsibility for production of forecasts was the same as that for the modelling of travel demand, as described in section 1.1.2. 2.1.2 The Use of the Models in Forecasting The models already described hâve been used to generate forecasts in the following way. Values of the exogenous, that is, externally determined, variables will be obtained, where available, from the prescribed scénarios. Where thèse are npt available, plausible assumptions about the évolution of thèse variables will hâve to be made. Total trip demand for the prescribed category can then be calculated. This demand is then assigned to geographical régions, using the given scénarios of population change, and our exogenous forecasts of the économie variables for the régions. In the second stage in some cases, demand arising in each région is divided into the three catégories, car, independent non-car and package non-car, according to the forecast économie characteristics of the région. In the third stage, the trips in each category are assigned to destination countries according to the proportions prévalent in 1977, and to zones within continental countries according to the proportions found in the 1971 surveys. For the forecasting of route choice, assumptions must be made about how journey characteristics change in the future, both with and without the link, and also how the parameters will change. We hâve assumed that value of time type parameters will change as personal income changes. Assumptions about surface and air transport costs are part of our prescribed scénarios. As for the parameters found in the impédance function for car travellers, the "disutility of distance" parameters are assumed to arise from two sources: an actual perceived cost of distance élément, which again will change according to our scénarios, and a "disutility of time" élément, which is assumed to be proportional to personal income. Changes in journey characteristics which will occur in the absence of the link hâve been incorporated by modifications to the network. Chief among them is the introduction of much faster train services between-Paris and the South of France, Switzerland and Italy. The effect of thèse planned improvements has been incorporated into the forecasts both with and without the link, so benefits.arising from them hâve not been attributed to the link. 085 2.2 Other technological developments are possible. Among those which might be considered are a major extension of jetfoil services. Another important change would be a réduction in air fares similar to those recently experienced on the North Atlantic. At the end of the report we recommend that sensitivity tests should be undertaken to examine thèse possibilities. When considering the addition of the link, there must of course also be added any associated improvements in surface transport on either side of the Channel. An important issue hère is whether thèse improvements, for example, fast rail links to Paris and Brussels, are accessible to travellers not U3ing the link. • The final issue to be considered hère is the possibility of traffic génération by the link. While our models include an influence of travel cost on total demand, the model structure is such that incorporation of the effect of the link at this stage would influence ail demand equally, including that which is not divertible to the link (e.g. car demand in the single-track tunnel case). We feel that the effect of the link in generating traffic will be strongly dépendent on the journey characteristics, and hâve estimated generated traffic separately for each origin-destination pair, and for each type of traffic. For each origin-destination flow, traffic is assumed to be generated by the réduction in impédance resulting from the présence of the link, and any concomitant fare réductions. The elasticity assumed is that found for the 1977 car travellers, naraely -3.08; in the absence of any direct évidence for non-car travellers, we hâve assumed that this elasticity applies to them also. 2.1.3 Developments in Passenger Traffic 1972-1977 An indication of the changes that hâve taken place in the passenger traffic between the UK and Continental Europe in the récent past can be obtained from the statistics published from the International Passenger Survey. Table 2.1.1 shows the traffic between Britain and the countries of Western Europe most likely to be affected by a fixed Channel crossing (the countries of the EEC, excluding the Irish Republic, plus Austria, Spain and Switzerland). A number of significant points émerge from thèse tables. Firstly, total traffic volumes hâve grown but erratically. The level of traffic fell between 1973 and 1974, but growth rates exceeded 10# between 1974 and 1975. The fall in traffic in 1974 can be explained by the 1973 oil price rise, and its repercussions on travel costs and économie growth rates. At the end of the period, traffic growth rates are lower than those experienced at the beginning of the period, but there is évidence of a new momentum in cross-Channel traffic in the last three years, with growth rates rising. Table 2.1.1 Visita Between the UK and Western Europe 1971 to 1977 (Thousands) UK Résidents Visiting Résidents of Western Europe Total Traffic Between UK Western Europe Visiting UK and Western Europe Year Air Travellers Sea Travellers Total Air Travellers Sea Travellers Total Air Travellers Sea Travellers Total 1971 3748 2365 6112 1408 1353 . 2761 5156 3718 8873 1972 4698 2472 7170 1509 1329 2838 6207 3801 10008 1973 5048 2516 7564 1648 1638 3286 6696 4154 10850 1974 4078 2750 6828 1806 1823 3629 5884 4573 10457 1975 4275 3389 7663 1826 . 2304 4129 6101 5693 11792 1976 4058 3071 7130 2066 2784 4851 6124 5855 11981 1977 3995 . 2971 6962 2320 3242 5561 6315 6213 12523 Traffic Growth Rates (,% per year) 1971 - 1972 25.3 4.5 17.3 7.2 -1.8 2.8 20.4 2.2 12.8 1972 - 1973 7.4 1.8 5.5 9.2 23.3 15.8 7.9 9-3 8.4 1973 - 1974 -19.2 9.3 -9.7 9.6 11.3 10.4 -12.1 10.1 -3.6 1974 - 1975 4.8 23.2 12.2 1.1 26.4 13.8 3.7 24.5 12.8 1975 - 1976 -5.1 -9.4 -7.0 13.1 20.3 17.5 0.4 2.8 1.6 1976 - 1977 -1.5 -3-3 -2.4 12.3 16.5 14.6 3.1 6.1 4.5 Annual Average 1971 to 1977 1.1 3.9 2.2 8.7 15.7 12.4 3.4 8.9 5.9 IV) 087 2.4 Secondly, sea traffic levels hâve grown at a faster rate than air traffic levels every year since 1972. As a resuit, the percentage of passengers crossing the Channel by sea has risen from 42$ in 1971, and a low point of 38$ in 1972 to 50$ in 1977. In large measure, thèse trends are a reflection of rising energy priées which pushed up air fares particularly sharply in 1974. In 1977, the number of cross-Channel visits made by air was still below the 1973 level. Thirdly, the more dynamic sector over the seven years has been traffic from the Continent to Britain, which grew 10$ per annura faster than traffic in the opposite direction. In 1971, résidents of the U.K. accounted for 69$ of cross-Channel passenger movements. This rosé to 72$ in 1972, but fell steadily to 56$ in 1977. Britain's slow économie growth rate, and the rising levels of real costs which Britain visitors encountered when travelling abroad, are undoubtedly the major factors behind this development. Il 088 (1 u (•] . 2.2 The Economie and Démographie Assumptions Mode i | We hâve forecast traffic crossing the Channel for two alternative . scénarios of économie development. The économie and démographie assumptions on [ ) which they are based hâve been agreed between the représentatives of the Commission and the consultants. ]) 2.2.1 Population Growth t J The population growth rates assumed are given in Table 2.2.1. They ' i are the same in both scénarios. 2.2.2 .Gross Domestic Product and Consumers' Expenditure [ . We présent forecasts for two scénarios of GDP growth, a 'low' scénario and a 'high1 one. The growth rates adopted are given in Table I i 2.2.2. A divergence between countries is envisagea in both scénarios over the period 1977-1985. Thereafter a convergence of growth rates is assumed for ail r- i . f countries. Consuraers* income and expenditure per head are assumed to change at ; •: the same rate as GDP per head in the low growth scénario, but more slowly in the high growth scénario. Thèse assumptions are suramarised in Table 2.2.3» Table 2.2.2 Proposed GDP Growth Rates {% P-a.) Country 1977 - 1985 1985 - 2000 Low High Low High Germany 1.6 3.0 2.90 3.5 France 2.0 3.7 2.0 3.5 Netherlands 1.7 3.1 2.0 3.5 Belgium/Luxembourg 1.3 2.4 2.0 3.5 Italy 1.3 2.4 2.0 3.5 Austria 2.0 3.8 2.0 3.5 Spain 2.3 4.3 2.0 3.5 Switzerland 1.4 2.7 2.0 3.5 United Kingdom 1.5 2.8 2.0 3.5 Others 1.7 3.2 2.0 3.5 Table 2.2.1 Population Levels (Thousands) and Growth Rates(% per year) Assuraed Scénarios Country 1985 Growth Rate 2000 Growth Rate Population 1977 - 1985 Population 1985 - 2000 Gerraany 60100 -0.3 57200 -0.3 France 55900 0.6 59000 0.4 Netherlands 14700 0.7 15600 0.4 Belgiura/Luxembourg 10400 0.3 10400 0.0 Italy 58800 0.5 61200 0.3 Austria 7700 0.3 8000 0.3 Spain 39100 0.8 45300 1.0 Switzerland 6900 7600 0.6 1.1 O United Kingdom 58600 0.6 60000 0.2 c© Other countries 0.4 0.3 to 090 2.7 Table 2.2.3 Consumers' Income and Expenditure Growth Rate (As % p.a. relative to GDP % p.a.) 1977 - 1985 1985 - 2000 Low High Low High Ail Countries 0.0 -0.3 00 -0.3 2.2.3 Changes in Relative Prices Changes in relative price levels can be separated into two effects. Pirstly, the gênerai price level in one country may rise or fall, relative to that in another. This effect is likely to hâve a strong influence on holiday-taking behaviour. Secondly, within a country, the relative prices of some goods will rise, while those of others fall. It is assumed under both our scénarios that any différences in the rates of inflation between ccimtries will be compensated for by adjustments in the exchange rates. Thus, the relative cost of visiting one country for résidents of another will not change in real terms. Changes in the real relative prices of certain goods will undoubtedly occur during the period covered by the study, and some of thèse changes will be of importance to our conclusions. The relative prices of the relevant goods will be influenced by the following factors:- costs of inputs; - costs of distribution; technical progress; - costs of labour; and - taxes and subsidies In addition, forecasts of "real" prices hâve to satisfy the constraint that some price index is constant. In principle, therefore, the price of a single class of goods cannot be defined as an élément in a scénario without référence to the prices of other goods, unless the demand for this good is very small compared to the total economy. We can, however, overcome this difficulty, by considering changes relative to the average in the economy. Of the influences on changes in relative prices, we consider the effect of differential technical progress, and that of the cost of fuel inputs, to be of greatest importance. 091 2.8 The effect of technical progress will be that, ail other things being equal, the priée of goods produced by sectors where technical progress is faster than average will tend to fall, whereas that of goods produced by sectors where technical progress is lower than average (usually labour-intensive sectors) will tend to rise. Investigations of past trends suggest that the priées of most manufactured goods are likely to fall with continued technical progress, while those of services, government, and housing are likely to rise. Superimposed on thèse changes arising from technical progress will be those arising from fuel price changes. It was felt that the effect of fuel priées was so prominent that their likely increase was discussed with officiais of the Commission, and it was agreed to associate a fuel price scénario with each GDP growth scénario. Under the 'low growth1 scénario, it is assumed that the relative cost of fuel will rise by 3% p.a. to 1985, and by 1.5$ p.a. thereafter. The assumption in the 'high growth' scénario is that the relative cost of energy will remain unchanged up tô 2000; in view of the considérable rise in the cost of energy experienced from 1977 to date, we hâve applied thèse rates of increase to the levels prevailing in July 1979i and hâve used the observed increase from 1977 up to that date. The assumptions underlying thèse scénarios were that the continued availability of relatively cheap fuel, at no higher cost than that at mid-1979, might be a prerequisite of a higher growth rate. We would argue that, given récent expérience, there is a case for testing the sensitivity of the results to an even higher rate of price increaae - say J>% to the end of the period. It is important to notice that the effect of this cannot simply be inferred by considering the proportion of a sector's costs which arise from fuel, and imposing a corresponding increase. The effect of "renormalising" priées, so that some priée index is constant, cannot be ignored. Thus, given a fuel-price increase, the priées of those goods for which the energy intensity is higher than that in the economy as a whole, will be higher than they would otherwise hâve been, and those of goods with a lower-than-average energy intensity will be lower than they would hâve been. 092 2.9 A detailed analysis of ail thèse factors is clearly outside the scope. of the présent study. In addition, certain trends, notably the cost of rail transport, appear to be in différent directions in différent countries. Given the difficulties, scénarios of the important price changes are set out in Table 2.2.4. They are mostly based on fuel intensity, although in the car manufacture and scheduled air transport sectors, faster-than-average technical progress has been allowed for. The scénarios hâve been agreed between the consultants, except that we feel doubtful in accepting the SNCF prédiction that rail costs will fall in real tenus over the study periods. 2.2.4 Other Exogenous Variables Variables other than those specified in the agreed scénarios which are required as input to the travel demand models are:- - household size; - friends and relatives abroad; - âge of completion of éducation; - average social grade; travel cost, by mode: car, rail, bus, ship and air; - car ovmership. We hâve attempted to use, as far as possible, accepted models or projections of thèse variables, and to establish consistency between the évolution of thèse variables and those specified in the agreed scénario. 2.2.5 Household Size The projections of the number of households in the U.K. hâve been published by the Department of the Environment (Housing Policy: Technical Volume, Part I: HMSO 1977). Thèse are given with the corresponding démographie projections, in Table 2.2.5. 0 33 2,10 Table 2.2.4 Price Scénarios (Jan. 79 price = 100) High Low 1985 2000 1985 2000 Sea transport -- passengers 106 106 109 113 m- freight 106 106 109 113 Port charges 100 100 100 100 Air transport -- scheduled 104. 5 96.9 111.0 11 1.0 - package 108. 0 108.0 117.4 126. 5 Rail passengers 100 100 100 100 Rail freight - UK 100 100 100 100 - continent 94. 1 84.1 95.6 88. 7 Road passengers UK 133 133 155 198 Road freight haulage 105 105 108 112 Cars 96 90 96 90 Table 2.2.5 POE Household and Population Projections . Average Average Number One-Person Number in Year Households Population in Household Households Multi-person Households 1971 16,779 55.6 3-31 3,330 3.89 1976 17,574 55.9 3.18 3,932 3.81 1981 18,254 56.3 3.08 4,497 3.77 1986 18,929 57.2 3.02 4,953 3.73 There is a substantiel décline projected in household size, most of which arises from an increase in the proportion of single-person households. We hâve used thèse projections of average household sizes although our démographie projections are différent from those used; after 1985 we assume the average household size remains constant. 094 2.11 Our projections of household size, population growth, and incorae growth enable us to infer scénarios of income per household and per person. Thèse are shown in Table 2.2.6. Table 2.2.6 Income Per Household and Per Person in the U.K. Growth of Income per Growth of Income per Person {% per year) Household (% per year) Scénario 1977-85 1985-2000 1977-85 1985-2000 High Growth 1.9 3.0 1.2 3.0 . Low Growth 0.9 1.8 0.2 1.8 2.2.6 Car Ownership To forecast car ownership levels in the UK, we used a model developed by Bâtes et al1 for the UK Department of Transport. This was a nodel of household car ownership and was estimated using time séries and cross-sectional data between 1969 and 1975. The explanatory variables were household income and car purchase costs. Réservations about the use of this model include:- (a) no dependence on running costs is included; and (b) the response to car purchase price is instantàneous, which is behaviourally implausible. Dealing with the first point, no influence of running costs was détectable during the estimation period, 1969-75, during which substantial changes in running costs did occur, but it has been suggested that inclusion of such a variable might give a better explanation of the rise in car ownership in 1976 and 1977, when real purchase priées were rising and real incorae was falling, but running costs were also falling. As for the second point, during the estimation period, the purohase price of cars fell steadily, so it was understandable that no lagged effect was détectable. We feel that thèse effects are an adéquate explanation of the fact that the 1977 car ownership figures are higher than forecast using the Bâtes model. Given that the effect of car purchase and income will be primarily on car purchase, we would expect the fall in car ownership predicted by the Bâtes model in 1975-77 actually to happen somewhat later. 1. Bâtes, J, Gunn, R. and Roberts, M., 1977: A disaggregate model of household car ownership: Departments of the Environment and Transport, Research Report 20. 095 2.12 The forecast car ownership levels per person, baaed on the Bâtes model and on the scénarios of household incorae and car purchase price given above, are shown in Table 2.2.7* Table 2.2.7 Forecasts of UK Car Ownership (cars per person) Scénario 1985 2000 High 0.262 0.367 Low 0.236 0.298 2.2.7 Travel Impédance in the UK This variable has two components, as described in section 1.2.2; the actual time taken, and the équivalent time of the costs incurred. For forecasting purposes we assume that there has been no change in average speeds, nor in the average distance travelled in the UK. The change in the cost per unit distance will be a weighted average of the changes, given in section 2.2.3 above, in the costs of car transport, other surface transport, and air transport. We hâve also incorporated a change in impédance for those changes in sea transport costs assumed to occur in our scénarios. 2.2.8 Friends and Relatives Abroad The period 1971-77 showed a very rapid rise in the number of UK résidents' visits to friends and relatives abroad, especially to those outside Europe. This suggests that the importance of this variable has increased. We hâve assumed that the proportion of households with friends or relatives abroad increases to 1.5 times its 1971 value by 1990 (as in the previous study) and continues to increase at the same rate. We assume that this increase applies equally to friends and relatives inside and outside Europe. 2.2.9 Hiscellaneous We believe that the explanatory power found in the cross-sectional analysis for the variables "social grade" and "âge of completion of éducation" is rather an indication of their rôle in representing "position" in society, and we do not believe that any changes in the values of thèse variables which will occur (eg, as more and more of the population has been educated up to 16 rather than 15) should be assumed to cause changes in travel behaviour. Accordingly, we hâve held the values of thèse explanatory variables constant throughout the forecast period. 096 2.13 2.3« The Base Forecast of UK Leisure Trips Abroad 2.3.1 The Base Forecast of Total Demand The logistic model described in Section 1.2.2 was fitted to the average propensity to travel abroad observed in 23 groupa of surveyed individuals, the individuals being grouped by the linear combination of socio-economic characteristics which best explained the variance in trip-making propensity. For forecasting purpose, it was assuraed that the distribution of values of the explanatory variables was such that the mean values for each of thèse groups would change by thé same proportion as the mean for thé population as a whole. Since the logistic équation was estimated on the mean properties of thèse groups, it was necessary to forecast propensities for similar groups, rather than for surveyed individuals or for the population as a whole. The résulta from this process are shown in Table 2.3«1« Thèse figurea hâve been adjusted for the small (1.5^) différence between the forecast for 1977 and the value observed in 1977; this différence is attributed to price changes and other effeota. Table 2.3.1 Forecaata of UK Leisure Tripa Abroad (thousanda of return trips) High Low 1977 1985 2000 1985 2000 Forecaat trips 7745 9344 14730 8444 1 1528 Implicit Annual 2.4* 3.1* 2.1JK Growth rate It is important to hâve some idea how sensitive thèse values are to the assumption made about saturation levels. The "beat" value of the saturation parameter in a leaat squarea aenae, ia 0.853; thia value was used to obtain the figures above. However, we tested différent modela, re-eatimating the other parametera, for a aaturation level fixed at 1.25 (close to that uaed in the previoua atudy), and for a aaturation level tending to infinity. Reaults for the high growth acenario are shown in Table 2.3.2. As can be seen, the model with aaturation level 1.25 ia nearly as good a fit as the be.st raodel,an d makea relatively little différence to the forecast. The model with no saturation gives a very much worae fit, (worse even than a straight line) and a aubstantially higher forecast. 097 2.14 Table 2.3«2 Sensitivity. of Forecasts to Saturation Assumptions High Growth Scénario Unexplained 1985-2000 Variance Assumption 1985 2000 Annual (1971) Trips Trips Growth Rate Saturation at 0.853 9344 14730 3.1 JE 0.7* Saturation at 1.25 9517 15386 3.2# 0.9# No saturation 9606 17590 4.1É 2.1* 2.3.2 Geographical Distribution of UK Demand We assume that the income growth described in the scénarios applied equally to ail the geographical divisions of the UK. The car ownership model used is estimated on national data and the authors explicitly atate that it should not be used to forecast changing car ownership patterns for régions within the UK. They do suggest that a similar functional form, with différent parameters, would be useful for this, but explicit models are not available. Given the absence of a suitable model, we are forced to assume that the UK growth rates of car ownership apply equally to ail geographical divisions. The interprétation of the geographical variable in the model of distribution over origins of UK demand is open to question, when used for forecasting purposes. It may be interpreted as purely a geographical effect, in which case there is no reason to forecast any change with time. On the other hand, "distance from Dover" may be a measurable proxy for "travel impédance to foreign destinations", in which case its effect on propensity to travel abroad might be expected to change. The question then arises of how to treat the observed flattening out of the distance effect beyond 160 miles. Given thèse difficulties, we hâve chosen to assume that the effect is purely geographical, and hence hâve forecast no change in its effect. We hâve no time séries of propensities to travel abroad by UK régions, so thèse hypothèses are not testable. 088 2.15 Given the assumption of uniform growth in car ownership and income, it would appear that our forecasts of propensity to travel abroad will change uniformly over the whole UK. The only contribution of differential holiday ï—> demand will be changes in the distribution of population about the UK. The Office of Population Censuses and Surveys (OPCS) publishes population projections for planning régions, in which the population growth between 1975 and 1991 was assumed to be confined to the South of England, the Midlands, East Anglia and Wales, with no growth in Scotland, Northern England, or Northern Ireland. Greater London is projected to décline in population. Our scénario of total population is différent from the OPCS projection, but we use the split of their projected population to infer growth rates consistent with our scénario. Thèse are shown in Table 2.3»3« J D99 2.16 Table 2.3«3 Forecast Population Growth by Région Share of Inferred Inferred UK Population Annual Population (Millions) Growth Rate 1977- 1985 Région 1977 1985 1991 1985 2000 1985 2000 Scotland 9-35? 9.2* 9.1* 5.4 5.4 0.5* 0.0* Iforth 5.6* 5.5* 5.4* 3.2 3.2 0.4* 0.0* NorthWest 11.6* 11.5* 11.4* 6.8 6.8 0.6* 0.0* Yorks & Humberside 8.8* 8.7* 8.6* 5.1 5.2 0.5* 0.1* East Midlands 6.6* 6.9* 7.0* 4.0 4.2 1.0* 0.3* West Midlands 9.3* 9.3* 9.3* 5.4 5.6 0.5* 0.2* Bast Anglia 3.2* 3.6* 3.7* 2.2 2.2 2.5* 0.0* Gréater London 12.5* 11.0* 10.5* 6.4 6.3 1.1* -0.1* Rest of South Eaat 17.7* 18.6* 19.1* 10.9 11.5 1.2* 0.4* South West 7.7* 8.0* 8.2* 4.7 4.9 1.1* 0.3* Wales 5.0* 5.0* 5.1* 2.9 3.1 0.4* 0.4* Northern Ireland 2.7* 2.7* 2.6* 1.6 1.6 0.8* 0.0* TOTAL UK 100.0* 100.0* 100.0* 58.6. 60.0 0.6* 0.2* The proportions derived from the OPCS projections for 1986 are assumed to apply for 1985t and those for 1991 are assumed to apply for future years. The numbers of leisure trips generated in each of thèse régions in the forecast years can then be calculated; thèse are then assigned to our zones on the assumption that the population of ail zones withia a planning région grow at the same rate. 2.3.3 Forecasts of UK Lsisure Trip Demand by Type The model of holiday type given in section 1.2 vas used to preduct the fraction of leisure trips from every UK zone in each of our three holiday catégories. Thèse fractions were then applied to the predicted UK demand for leisure trips abroad by zone. The changes in the proportion of leisure trips abroad going by car follow closely thé car ownership scénario described above. In both scénarios this fraction is substantially higher by 2000 than at présent. The split of non-car trips betveen independent and package trips remains roughly constant, the variation being due to changes in the geographical distribution of population. The forecast split between the three catégories is shovn in Table 2.3*4. 100 2.17 Table 2.3.4 Forecasts of the Percentage Split by Type of UK Leisure Trips High Low 1977 1885 2000 1985 2000 Car 12.4 12.8 20. 6 10.9 15.5 Independent 43.1 40.4 36. 7 41.3 39.1 Package 44.5 46.8 42. 6 47.8 45.3 Forecasts of U.K. Leisure Trip Demand by Destination As described in section 1.2, the main influence on the destination choice of U.K. leisure travellers appears to be relative costs of living. As raentioned above, the agreed scénario has relative costs of living unchanged. Our forecasts of destination choice within each category of traveller will therefore be of unchanged proportions. For car travellers, we hâve found a significant dependence of destination choice on travel impédance, using 1977 régional data. The effects of this would be fairly small for the car forecasts, as far as total numbers to a given country are concernée!. Since the correct interprétation of this finding in the context of our "top-down" model structure is not clear, and we were unable to find a similar relationship for non-car travellers, we hâve not taken this into account in the without-link forecasts. We hâve, however, used the elasticity thus determined to estimate generated traffic, as described below. 101 !j 2.4 The Base Forecast of Continental Leisure Trips to the UK i > (Summary of SETEC Rapport de Phase 3, sections 4.2 and 5.1) v~~ The proportions of total leisure travel originating in each country II are not expected to hâve varied greatly between 1977 and 2000 in either the !• j low or high growth case. j"J In 1977 27^ were packaged trips, 46% independent and 26% were car-accompanied. Before allowing for any traffic generated by a fixed link, -• the proportions are predicted to change to the following by 2000. Package Independent Car-Accompanied i i Low growth 12 43 34 High growth 18 44 38 [ | As may be seen the major change is from package to car-accompanied and this is strongly related to growing incomes. 102 2.19 2.5 Summary of Base Forecasts of Leisure Trips The forecasts are brought together in Table 2.5.1. In 1977 51$ of leisure trips across the channel were British. By 2000 the proportion is expected to hâve fallen to 40% in the low growth scénario or 33$ in the high growth. While the propensity of someone from the UK of a given income group to travel to Continental Europe will remain higher than that of anyone from a similar income group from a Continental country, the growth of Continental incomes relative to those of the UK will mean that by 1985 the number of Continental travellers coming to Britain will exceed UK leisure passengers to the Continent. By the end of the century on the high growth scénario 11.5 million from the UK will be going to the Continent while nearly 20 million from the Continent and 3»5 million from the rest of the world will be coming to Europe. The proportions and absolute numbers would alter of course if UK incorae growth increased relative to that of other countries. One would expect more travellers to the Continent and fewer from it. While independent travellers may be increasingly attracted to destinations further from their home countries, this will be offset by rising energy costs and the inconvenience of driving far to the increasing proportion of car accompanied passengers. The absolute increase in the number of tourists is considérable: in the UK from just above 6 million in 1977 to 9.5 million in 1985 and 23 million in 2000 on the high growth scénario; but it should be remembered that growing affluence does lead to more tourism and that an increasing proportion of thèse journeys will not be in the summer peak as second and third holidays become more comraon. The growth is less dramatic in the low growth case: frora 6 million in 1977 to 7.5 million in 1985 and 13 million in 2000. 103 2.20 Table 2,5.1 Summary of Demand Forecasts Low growth High growth 1985 2000 1985 2000 UK résidents: leisure - with a car 921 1795 1200 3054 - without a car 6057 . 7835 6555 9404 UK résidents: business - with a car 147 213 715 335 - without a car 1292 1831 1537 2910 Continental résidents: leisure - with a car or coach 2466 . . 4769 3130 8732 - oii foot 3574 6298 4383 11117 Continental résidents: business - with a car or coach 107 204 136 401 - on foot 1204 2238 1513 4396 Non-European résidents 1623 . 2120 1960 3400 Notes: (a) Units are thousands of return trips; (b) "UK résidents: leisure" includes travel to ail Continental European destinations; (c) "UK résidents: business" includes travel to ail destinations outside the UK; (d) "Continental résidents" includes travellers résident in the countries of the EEC (excluding the Irish Republic), Switzerland, Austria, Spain, Portugal, Greece and Yugoslavia. 104 2.6 The Base Forecast of the Demand for Business Trips 2.6.1 Business Trips by UK Résidents The model of UK résidents1 business trips to continental Europe is formulated as described in section 1.4 with the equation:- where Tj is the number of return business trips, in thousands, raadet o country j by U.K. résidents; aj is a country-specific attraction factor; Fj is the scheduled air fare from Heathrow to an appropriate commercial centre in country j; Gj is the value of the annual gross domestic product of country j, expressed in billions of 1975 US$; Xj is the value of the annual exports from the U.K., to country j, expressed in billions of 1975 US$. Forecasts of the explanatory variable Xj are given by SETEC, and are summarised in Table 2.6.1. Forecasts of GDP and of air fare changes are given above. The resulting forecasts of business trips are given in Table 2.6.2. l.J 105 2.22 Table 2.6.1 Forecasts of Explanatory Variables for Business Travel î) Demand Model 1985 2000 Scénario Country Exports Exports GDP From UK GD? From UK 0 (1977=100) (1977=100) (1977=100) (1977=100) France 133.7 171.8 224.0 448.2 Belgium & Lux 120.9 155.8 202.5 - 487.4 Netherlands 127.7 158.2 213-9 418.4 High Growth W. Germany 126.7 170.5 212.3 518.9 Italy 132.7 182.4 222.3 562.6 Switzerland 123.7 127.7 207.2 277.0 Austria 134.8 143.3 225.8 285.4 Spain 140.0 143.3 234.4 265.6 Rest of World 128.7 - 139.0 215.6 305.9 France 117.2 139.0 157.7 250.3 Belgium. & Lux 110.9 131.7 149.3 262.3 Netherlands 114.4 132.7 154.0 239.0 W. Germany 113.5 139.0 152.8 272.8 Low Growth Italy 116.3 142.2 156.5 283-2 Switzerland 118.8 114.4 150.5 178.3 Austria 117.2 122.8 157.7 183.1 Spain 120.0 121.8 161.5 173.9 Rest of World 144.4 120.9 154.0 191.1 ( I •• Source: SETEC Economie 106 2.23 Table 2.6.2 Forecasts of U.K. Business Travel Demand (Thousands of Return Trips) High Growth Low Growth Country 1977 1985 2000 1985 2000 France 306 439 803 370 523 Belgium & Luxembourg 154 206 417 180 266 Netherlands 179 245 452 211 298 West Germany 280 398 789 337 495 Italy 83 129 269 101 150 Switzerland 60 76 130 66 86 Austria 19 27 45 22 28 S pain 60 85 139 69 87 Sub-total 1,141 1,605 3,044 1,356 1,933 Rest of World 863 1,253 2,348 972 1,289 TOTAL 2,004 2,858 5,392 2,328 3,222 107 2.6.2 Business Trips by Continental Résidents (Sumraary of SETEC Rapport, Phase 3, section 5.12) For business trips originating in Continental Europe, the changes in the proportions originating in each couritry will be greater than for leisure travel. The French proportion is forecast to be almost constant. The proportions of Gerraans and Italians are forecast to increase substantially, especially the first, while those from the Netherlands, Switzerland and Spain will fall. Thèse trends will be less marked in the low growth case. 2.6.3 Summary of the Forecasts of Business Trips The consultants' separate forecasts of business trips are brought together and summarised in Table 2.6.3, which also shows SETEC's forecast of how many in each category travel with a car. The proportion of British business travellers abroad falls sharply frora just under 50% in 1977 to 42$ in 2000 on the low growth scénario and 37$ on the high growth scénario. This is because our research showed that the volume of business travel i'rom the UK is strongly relateà to its exports; and on the basis of the économie growth assumption adopted hère, UK imports from Europe increase much faster than UK exports to Europe. If UK exports were to increase at a faster rate, so would its business travel. The proportion of business travel by car is not expected to vary much on either scénario. Business travel by car is about 5 to 6% of passengers but only 5% of vehicles. 108 2.25 Table 2.6.3 Forecasts of the Demand for Business Trips (thousands of return trips) 1985 2000 Catégories of 1977 High growth Low growth High growth Low growth Traffic scénario scénario scénario scénario l.UK Without a car 1O9Ô: 1537 1292 2910 1831 Car-accompanied 123.- 175 147 335 213 Number of vehicles 59 83 70 160 101 accompanied 2.CONTINENTAL Without a car 912: 1513 1204 4396 2238 Car-accompanied 81 136 107 401 204 Number of vehicles 39 65 51 191 97 accompanied 3.OUTSIDE EUROPE Without a car 271 354 301 617 392 Car-accompanied 4 5 4 9 5 Number of vehicles 2 2 2 4 2 accompanied i TOTAL (1+2+3) 2481 3720 3055 8668 4883 ! TOTAL: UK 1213 1712 1439 3245 2044 ! (%) 48.9 46.0 47.1 37.4 41.9 j Continental 993 1649 1311 4797 2422 ! (%) 40.0 44.3 42.9 55.3 50.0 ! Outside Europe 275 359 305 626 397 | (%) 11.1 9.7 10.0 7.2 8.1 ! i (Without a car 2273 3404 2797 7923 4461 i ( (%) 91.6 91.5 91.6 91.4 91.4 TOTAL(Car- ( accompanied 208 316 258 745 422 (%) 8.4 8.5 8.4 8.6 8.6 TOTAL VEHICLES 100 150 123 355 200 109 2.26 2.7 The Base Forecàst of the Demand for Travel by Non-Europeans The raodel for thèse travellers is set out in section 1.5; the model expresses the demand for travel to Europe in terms of the population of the overseas région, its GDP per head, and the cost of travel to Europe. Our scénarios give a uniform growth rate of population and GDP for ail countries outside Europe. We will assume that thèse travellers get to Europe by air, and that the price of air travel moves uniformly for ail such travellers, as described above. The growth will therefore be the same for ail catégories of non-European travellers, and we will assume that the number of trips between the U.K. and continental Europe generated by each traveller remains constant. The results of this are shown in Tables 2.7.1. Table 2.7.1 Forecasts of Non-European Trips between UK and Continental Europe (thpusands of single trips) High Growth Low Growth Country 1985 2000 185 2000 France 624 1,083 517 675 Belgium & Luxembourg 203 353 168 220 Netherlands 393 682 326 426 West Gerraany 407 706 337 440 Italy 760 1,318 629 822 Denmark 203 353 168 220 Switzerland 258 447 213 279 Austria 149 259 124 161 Spain 149 259 124 161 Portugal 27 47 . 22 29 Greece 326 565 270 352 Yugoslavia 27 47 22 29 Norway, Sweden & Finland 393 682 326 426 3,919 6,801 3,246 4,250 110 2.27 2.8 The Base Forecast for Freight Traffic The base forecast of freight traffic is carried out by:- (a) forecasting total trade between the UK and continental countries; and then (b) forecasting the "unitised" élément of thèse total trade flows. 2.8.1 Total Trade Flows SETEC hâve produced forecasts of trade between the UK and continental countries by commodity, by country, by import/export for both of the scénarios for the years 1985 and 2000. Thèse are expressed in tonnage terras, net of packaging. The base years for thèse forecasts is 1976. 2.8.2 Unitised Trade Flows A fixed link will carry only unitised traffic, that is, ro-ro container or rail wagon, as appropriate to. the nature of the link. The potential trade divertible to a fixed link is therefore trade flows that are expected to be unitised. In the previous Channel tunnel studies, a detailed analysis was raade of the possibility of the existence of a fixed link giving rise to induced unitisation of traffic. The conclusion of this exercise was this would be minimal, and we hâve assumed it to be nil in this current study. Unitised trade flows were forecast by the application of forecast unitisation rates to SETEC's forecast commodity flows. The levels of unitisation in the base year were derived. SETEC's base year was 1976. The detailed unitised flows derived for the calibration of the freight route choice model used 1 977 as the base year. In order to dérive base year unitisation rates the coramon year of 1976 was adopted. Estimâtes of the unitised trade flows for 1 976 were calculated by applying the overall change in trade between the UK and the countries considered, as given by HMC & E data, to the unitised flows derived for 1977. Thèse conversion factors are given at Appendix G. This implicitly assumes no change occurred in the overall levels of unitisation between the two years for trade between the UK and each continental country. 111 2.28 We therefore had for 1976, for each country, and for imports and exports:- (a) an estimate of unitised trade, over ail commodities, expressed in NPC tonnage terms, i.e. inclusive of packaging; and (b) the total trade flows by commodity, expressed in net tonnage tenns. In addition, the HM C & E data for 1 977 provided estimâtes of the pénétration of unitiaation at the broad commodity classification of:- (a) foodstuffs; (b) manufactured good; and (c) basic materials; which therefore provided an apprôximate commodity breakdown of the unitised flows, after allowing for a packaging factor. We also had access to the working papers of a detailed NPC study^ which analysed unitisation rates at a more detailed commodity level. With this information we were able to dérive the unitisation rates that existed in the base year. Thèse unitisation rates, which are presented at Appendix G, represent a 'composite rate' in that the single rate allows for both the "true" level of unitisation and also an . allowance for packaging. Thus, if out of a total flow of 100 tonnes, excluding packaging, the unitised trade was 55 tonnes, including packaging, a rate of 55$ was derived which would represent, say, a true rate of 50$, plus, say, a 10$ allowance for packaging. It was assumed that the discrepancy in trade in basic commodities that exista between the HM C S E and the OECD figures, as used by SETEC, relates to non-unitised trade. Therefore, our derived unitisation rate, relative to the HM C & E estimate of the pénétration of unitisation, reflects the use of a différent base tonnage figure. 2.8.3 The Forecast of Unitisation Having derived the unitisation rates for the base year of 1976, it was necessary to forecast the values for the years 1985 and 2000. This had regard to the NPC Report No. 11 and discussions with NPC personnel. Their opinion was that the considérable increase in pénétration over the past décade 1. NPC Report No. 11 : GB Non-Fuel Traffic: Forecasts for 1980 and 1985 (November 1977) 112 2.29 is now approaching a plateau. It was considered therefore that a conservative .reasonable estimate of the future unitisation rates would be:- (a) an increase of two percentage points for each commodity group by the year 1985; (b) a further increase of two percentage points for each commodity group by the year 200. In view of the uncertainty surrounding the detailed commodity unitisation rates for 1976, thèse assumptions were applied at the three broad commodity classifications. In addition to the countries for which SETEC provided detailed commodity forecasts, it was necessary to include also:- (a) forecasts of Irish unitised trade with continental countries that used Great Britain as a land-bridge; (b) forecasts of unitised trade to countries in the secondary zone; and (c) forecasts of entrepôt traffic, i.e. unitised trade between the UK and other countries in the world that is routed through continental ports. With respect to Irish traffic, SETEC provided forecast of trade between Ireland and ths EEC (other than the UK). It was assumed that the unitised élément of such trade via the UK would grow at the same rate as overall trade, while allowing for a small growth in unitisation rates. Unitised trade between the UK and Eastern European countries not included in the SETEC forecast was assumed to grow at the same rate as unitised trade with the SETEC forecast area. Porecasts of unitised entrepôt trade were provided explicitly by SETEC. The overall forecasts of unitised trade are shown below in Tables 2.8.1 and 2.8.2. It is assumed that the UK régional .breakdown for trade with each continental country is the same as in the base year, 1977, and the régional breakdown within each continental country, where appropriate, is also the same as that assumed for the base year. In the absence of better information, Irish traffic via the UK is assumed to hâve the same continental origin or destination as ail UK traffic, excluding entrepôt. The UK origin or destination of such traffic is assumed to be Wales and North West of England. 113 2.30 Table 2.8.1 : Forecasts of Unitised Trade ; Imports (000 tonnes) 1985 2000 COUNTRY 1977 LOW HIGH LOW HIGH Belgium/Luxembourg 1,233 1,668 1,828 2,517 3,762 Denmark 737 835 896 1,051 1,268 France 2,140 2,517 2,862 3,409 4,583 Italy 1,060 - 1,455 1,683 2,487 4,237 Netherlands 1,618 2,634 2,921 3,993 5,294 West Germany 2,160 2,624 3,055 4,243 6,624 Total EEC 8,948 11,733 13,245 17,700 25,768 Austria 252 317 345 393 551 Switzerland 157 176 182 239 329 Spain 494 582 660 741 954 Total Primary 9,851 12,808 14,432 19,073 27,602 Secondary Zone 371 511 585 771 1,126 Entrepôt Trade:- Belgium 161 182 201 256 365 France 33 38 42 53 76 Netherlands 412 468 518 658 942 606 688 761 967 1,383 Irish via UK 269 342 398 514 777 Total 11,097 14,349 16,175 21,323 30,889 Index 100.0 129.3 145.8 192.2 278.4 lift 2.31 Table 2.8.2; Forecasts of Unitised Trade : Exports (000 tonnes) 1985 2000 COUNTRY 1977 LOW HIGH LOW HIGH Belgium/Luxembourg 953 1,197 1,449 2,447 4,063 Denmark 482 600 709 1,029 1,672 France 1,330 1,745 2,162 3,054 5,132 Italy 636 887 1,119 1,323 2,477 Netherlands 1,124 1,834 2,153 3,163 4,853 West Germany 1,315 1,611 1,963 3,156 5,625 Total EEC 5,840 7,874 9,555 14,172 23,822 Austria 234 241 280 339 548 Switzerland 216 200 218 316 478 Spain 222 205 237 269 401 Total Primary Zone 6,812 8,520 10,290 15,096 25,249 Secondary Zone 364 476 573 689 1,151 Entrepôt Trade:-' Belgium 289 331 366 467 665 France 311 356 394 502 716 Netherlands ' 506 579 641 816 1,164 1,106 1,266 1,401 1,785 2,545 Irish via UK 219 267 307 381 508 Total 8,201 10,531 12,571 17,951 29,452 Index 100.0 128.4 153.3 218.9 359.1 115 ROUTE CHOICE IN THE ABSENCE OF A LINK 116 3.1 3» 1 Croaa-Channel Services in the Absence of a Link The récent rapid developments in the nuraber and variety of cross-Channel services available give an indication of the risk in trying to forecast what services will be available to the leisure passenger in 1985 and 2000, even in the absence of a new fixed link. For our forecasts, we hâve taken the conservative assumption that any new services offered can be regarded as providing a similar service to some existing ferry. It is worth noting that had this assumption been made in 1973, the advent of the Sheerness-Vlissingen service and the ferries to Brittany ports would hâve made the assumption invalid. However. the présent coverage of South and East coast ports is now so complète that it is difficult to see how a new service could be introduced that offers a substantially différent alternative, appreciably improving journey times to any destinations. The obvious exception to this is the possibility of jetfoil services becoming established. Bven if they operate from ports at or near those presently in use, as does the Brighton-Dieppe jetfoil, they offer a service which is substantially faster than the comparable ship. However, more importantly, jetfoil services may soon operate rsgularly from a port in Central London, which would make a radical différence to the range of services offered. A related problem is the compétition from air services. Again, we hâve adopted a conservative position, assuming no change in air transit times, and fare changes based solely on technological changes and on fuel prices. However, it is widely believed that the short-haul European air travel market is on the verge of two dramatic changes. Firstly, it is likely that the London to Paris, Brussels and Amsterdam services will in the near future adopt a "shuttle" System, which will considerably diminish presently unavoidable waiting time. This change has had important effects on the competitiveness of air services vis-a-vis rail on such routes as London-Glasgow. Secondly, there has recently been much spéculation that the dramatic fare réductions recently observed on transatlantic air routes could occur on European routes, following increased compétition from small, independent airlines and the résultant breakdown of the pricing cartel. Suggested fare réductions hâve been as great as 60%, This development would clearly hâve a great effect on the potentiel demand for a link, at least from passengers without cars. However, it is questionable to what extent air traffic on thèse routes would then be constrained by airport capacity. It has not been within our terms of référence to consider what pattern. or amount of airport development might be optimal in this respect. 117 3.2 We hâve assumed that car ferry services will exist offering:- (a) direct services from Harwich and Newcastle to Norway and Denmark; (b) direct services from Harwich or Pelixstowe to Sweden and North Germany; (c) a direct service to Spain - we hâve assumed it opérâtes out of Plymouth, as in 1979, rather than Southampton as in 19*77; (d) services from Hull to Rotterdam and Zeebrugge; (e) services from Harwich or Felixstowe to Belgium and Netherlands ports (East Coast); (f) Sheerness-Vlissingen; (g) Dover, Folkstone or Ramsgate to Ostend or Zeebrugge (Belgian Straits); (h) Dover, Folkstone or Ramsgate to Dunkerque, Calais or Boulogne (French Straits); (i) Newhaven-Dieppe; (j) Southampton, Portsraouth or Weymouth to Le Havre or Cherbourg (South Coast); (k) Portsmouth or Plymouth to ports in Brittany. For non-car travellers, the choice is less critical since the more peripheral routes are less patronised by thèse travellers. We hâve simply assumed services comparable to those in 1977 operate. No new air services hâve been introduced. We hâve also assumed that présent fare relativities between the ferries are maintained, with one exception. Hovercraft services are at présent offering a speedier service at priées roughly the same as the corresponding ship services, and are making a loss or at best a poor .return on capital. It seemed unreasonable to postulate increasing traffic with investment necessary to replace existing craft for a sector in such a position. Accordingly, we assumed that the hovercraft services would charge a premium fare for car travellers, namely 110$ of the corresponding ship fare, except in 1985 in the low growth scénario where the factor was 105$. h us 3.3 fl i. J l-> Certain very récent developments may appear to call into question , the validity of some of our assumptions about the ferry services. Pirstly, the [j tariffs announced by the operators for the 1980 season appear to indicate at least a partial breakdown of their cartel, with conséquent price réductions; this aspect has been heavily emphasised in their advertising. However, doser examination of the tariffs shows that the increased compétition is largely | } confined to off-peak travel, with the peak rates increasing in a similar fashion to that forecast. Also, the price war may be a transitory effect f ] caused by the introduction of a nuraber of new vessels at the same time, by »•••' both Townsend-Thoresen and Sealink. A second and rather more serious ,-.-, developraent is the claim by Townsend-Thoresen to be able to eut the t ; . Dover-Calais crosing time by 15 minutes, and to eut waiting and clearance time by "up to" 30 minutes by improved loading and unloading procédures on their ( , new vessels. Thèse improvements, if sustainable, would certainly improve the competitiveness of the ferries vis-a-vis a new link, but we feel that the ' increased speeds may well incur substantially higher fuel costs, and thence a more rapid rate of fare increase. Thèse developments may merit further ! "• considération. 119 3.4 3.2 UK Leisure Traffic 3*2.1 Car Travellers The patterns of route choice predicted for our two scénarios in the absence of a link are shown in Table 3.2.1. The changes in exogenous circumstances which bring about changes in route choice patterns are:- (a) changes in relative costs of road travel and sea travel; and (b) changes in the relative disutility of money and time. In the low growth scénario, both ferry costs and road costs increase, as a resuit of increasing fuel costs, but the increase in road costs is slightly higher. This will hâve the effect of transferring traffic from the shorter sea routes to the longer, which generally involve less road travel. The increase in value of time will reinforce this change, since the disutility of time on the road is greater than that on the ferry, and the longer ferries hâve higher fares. In the high growth scénario, the first effect does not occur, but the second effect is stronger. Thus in both scénarios we see that the increase in traffic on the shorter routes, the French and Belgian Straits, is less than that on the longer routes. Thèse changes are a continuation of the trends observed in the récent past. 3.2.2 Independent Non-Car Travellers Table 3.2.1 also shows the patterns of route choice predicted for independent non-car travellers. For thèse travellers, the principal factors bringing about changes in route choice patterns are:- (a) changes in the relative costs of air and surface transport; and (b) changes in the relative disutility of time and money. In the low growth scénario, air fares are forecast^to rise faster than those for surface transport because of higher energy costs. This will tend to favour surface modes. However, the increased value of time will more than counteract this effect, and we find that the increase in traffic in this category is raostly air traffic. In the high growth scénario, the switch to air is accentuated by a substantial fall in air fares since in this scénario, the faster-than-average technical progress assuraed for air travel has not been compensated for by increased fuel priées. This causes a large décline in the numbers in this category using surface transport, the numbers becoming negligible in the year 2000 on ail but the most central routes. Within the air travellers, there is a slightly increased tendency to use régional airports, largely because of the relative décline in London's population. Table 3-2.1: Route choioe for UK leieure travellers in the absence of a fixed link Car travellers Independent non-car travellers Package non-car travellers Crossing High Low Higl1 High Group Low Low 1985 2000 1985 2000 1985 2000 1985 2000 1985 2000 1985 2000 Norway & Sveden direot 24. 4 47.8 31.5 81. 3 » * » * * « * • Denmark direct 23. 9 46.5 31.3 79. 3 11.3 7.1 9-1 2.0 3-4 2.9 2.4 0.7 Germany direct 7.7 23.4 8.6 30. 7 3-4 1.1 1.8 0.0 0.7 0.5 0.4 0.0 Spain direot 16. 1 42.9 18.1 59.9 0.1 0.0 0.0 0.0 0.2 0.0 0.1 0.0 Hull routes 28. 4 114-3 36.8 175.0 3-8 0.9 1.6 0.1 1.7 1.6 0.7 0.1 Haven-Belg, Netherlands 82. 6 148.6 110.0 260. 5 94-5 55.7 61.0 11.8 35.4 30.2 21.2 7.5 Sheernees-Vlissingen 18. 1 44-1 25.8 58. 8 7.1 3.4 4.1 0.3 9.0 2.8 1-9 0.3 Belgian Straits 88. 2 161.0 113.6 208. 0 194.8 180.0 165-6 105.2 90.4 98.1 72.7 55.1 Prench Straits (ehip) 230. 3 355 1 331.8 693 9 837.1 902.7 758 6 4500 399-3 487.3 317.7 229.0 n " (hover) 47.4 27-5 18.6 72. 8 Newhavén-Di eppe 128. 1 308.4 181.7 542.6 94.2 71.8 71.8 22.4 86.6 87.6 57-6 23.7 South Coast 155-9 288.2 204.7 557.0 48.0 38.5 37.3 18.8 61.4 61.2 40.4 17.5 Brittany 69. 5 187-4 87.4 234.4 8.1 4.2 5-1 1.1 4-6 3-7 2.6 0.9 Total surface 920. 6 1795.2 1199.9 3054.2 13Q2.7 1266.3. 1116.0 611.7 692.7 775-9 517.7 334.8 Air-via London 1017.8 1661.5 1345-7 2725.0 2443.1 3267.3 2835-1 4404.1 Air-not via London 222.2 358.5 290.4 603.7 379.4 504.0 450.1 725.1 Total Air 1240.0 2021.0 1636.1 3328.7 2822.5 377L3 3285.2 5129.2 Total 2542.7 3287.3 2752-1 3940.4 3515.2 4547.2 3802.9 5464.0 Notes: 1. Destinations outside Europe exoluded.. 2. Unité are thousands of return trips. 3. "Haven-Belg, Neth" comprises crossings from Harwich and Pelixetowe to ports in Belgium and the Netherlands. 4. "South Coast" comprises crossings fin m Southampton, Portsmouth and Weymouth to LeHavre and Cherbourg. 5. "Brittany" comprises crossings from British ports to St Malo and Roscoff. 6. * indicatea that trafflc on thèse routes was not considered in the route choice model. V VJl " 121 ,. j r '• ' . J 3.2.3 Package Travellers ' • For thèse travellers, the swing to air travel is not so marked as "^-: for independent travellers, as our scénarios postulate limited scope for technical improvements and hence for further falls price for package air : i travel. Nevertheless, the same conclusion applies, that compétition from air will reduce travellers on certain routes to negligible proportions by 2000. i This projection probably represents too strong a conclusion, since the operators of thèse routes themselves offer a large number of packages. : However, most of thèse, in récent years, hâve been aimed at car travellers. 122 3.7 3.6 Freight Route Choice The results of the route choice forecasts in the absence of the link are shown below in Table 3«6.1. Thèse indicate the influence of the price scénarios. The increase in the share of rail wagon traffic reflects the decrease in the relative price of rail haulage, whilst the relative price of other modes is increasing, as described in section 2.2.3» This is raost marked in the 'low GDP' growth scénario. Table 3«6.1 Route Choice Model Results; No-Link Situation (thousands of tonnes, import & exports combined) 1985 2000 1977 LOW HIGH LOW HIGH Ro-ro -13,349 13,756 17,287 18,438 32,601 Container 5,105 6,665 7,553 9,656 15,320 Rail wagon 855 4,417 3,856 11,107 12,303 19,309 24,837 28,696 39,201 60,224